Not sure which lesson to do first? Find your level and get a plan →

Math · Statistics & Probability · Grade 6-8

Randomness

Randomness is the property that individual outcomes can't be predicted, while over many trials they settle into stable proportions.

Venn diagram of Event A and Event BA two-event view of randomness.ABA ∩ B
A: Event A
B: Event B
A two-event view of randomness.

Quick Answer

Randomness is the property that individual outcomes can't be predicted, while over many trials they settle into stable proportions. Coin flips, dice rolls, shuffled cards and lottery draws all behave this way: you cannot call the next result, but you can say what share of results to expect over hundreds of tries. Use the idea to set fair expectations, to recognise that long runs of one result happen naturally, and to tell genuine chance apart from a process that leans one way.

Why This Matters

Randomness explains why a fair coin can land heads five times in a row with nothing wrong, and why that tells you nothing about the sixth flip. It is also a tool: random selection is how surveys and experiments stay fair. Without the idea, people see patterns in runs of heads, bet on outcomes being 'due', and miss real bias.

Intuitive Explanation

Shake a jar holding 30 red and 70 blue beads, then pull one out without looking. Will it be red? You cannot know. Put it back, shake, draw again, and keep going. After 10 draws you might have 1 red or 5. After 1,000 draws, the share of reds will almost certainly be close to 30%. That is randomness: no pattern in any single draw, a steady proportion in the long run.

Random sequences contain longer runs than people expect. Ask someone to invent 20 coin flips and they usually switch between heads and tails a lot, avoiding runs like HHHH. Real flips are different: about 3 in 4 sets of 20 fair flips contain a run of 4 or more. Each flip has no memory of the last one, so after four heads the chance of heads on the fifth is still exactly one half.

Randomness is not the same as bias. A coin that lands heads 90 times in 100 has a lean you can predict and bet on, which is the opposite of random. Random also does not mean that anything goes: a fair die is unpredictable roll by roll, yet over 600 rolls each face shows up close to 100 times.

Think of it like this. Popcorn in a pan behaves a bit like a random process. You cannot tell which kernel will pop next, yet you can be confident that nearly all of them will have popped after a few minutes. Single events are unpredictable while the overall pattern is steady. The picture breaks in one place: kernels heat up as time passes, while a coin flip does not depend on the flips before it.

Core idea. Randomness means each single outcome is unpredictable, yet many repetitions follow precise probability rules.


When to Use

Think randomness when a process gives unpredictable single results but stable long-run shares: flips, rolls, shuffles, random draws, random selection for a survey. It is also the idea to bring when someone says an outcome is due after a run of the other result. If one outcome comes up far more often than the setup allows, suspect bias instead; if you need a number for how likely an outcome is, that is probability.

Tip. Ask: Is each single outcome unpredictable while the long-run rates stay fixed?

How to Recognize It

Before you call a process random, look at how its results are produced, such as a fair coin, a shuffle or a computer draw, and at whether some results come up far more often than the setup allows.

  1. Over many trials, does one outcome turn up far more often than the setup says it should?

    If yes, suspect bias, a steady tilt: a die showing 6 in 300 of 600 rolls is loaded, since a fair die gives about 100. Shares near the predicted ones fit randomness, even when short runs look odd.

  2. Does the question ask for a number, such as the chance of rolling a 3?

    If yes, that is probability, here 16. Randomness is the reason a number is needed at all; the probability is the number itself.

  3. Is someone reading meaning into a run, such as six tails in a row, or saying a result is due?

    If yes, randomness is the idea that answers them: runs like that happen naturally, and a coin has no memory of past flips. The next flip is still a 50-50 chance.

  4. Is the randomness a small wobble around a value you are trying to measure?

    Then the unpredictable part is noise, the scatter around a real signal. Randomness is the broader idea: it also covers processes like dice, where the outcome itself is the whole point.

Randomness compared with Probability, Noise, and Bias / non-random

Randomness, probability, noise and bias all come up whenever results vary, so a lopsided run can be blamed on any of them. Ask whether you are describing an unpredictable process, putting a number on one outcome, measuring wobble around a true value, or spotting a steady tilt.

Randomness

Meaning
Use this when individual outcomes are unpredictable but the long-run proportions are governed by fixed probabilities.
Key test
Is each single outcome unpredictable while the long-run rates stay fixed?
Example
A fair coin lands H,H,H,H. What's the chance the next flip is heads?

Probability

Meaning
Is the number measuring how likely an outcome is, not the unpredictability itself.
Key test
Use when you need to quantify the likelihood, not describe the process.
Formula
P(E)=favorabletotal
Example
P(red)=35

Noise

Meaning
Is random variation around a known signal, not pure outcome unpredictability.
Key test
Use when there's an underlying pattern with random wiggle on top.
Example
Static around a song

Bias / non-random

Meaning
Is a systematic, predictable tilt, the opposite of randomness.
Key test
Use when one outcome is consistently favored by the method.
Example
A weighted die landing on 6 too often

Worked Examples

Example 1 — Predicting a fair coin

Easy

A fair coin lands H,H,H,H. What's the chance the next flip is heads?
Show the steps and the answer
  1. Note what is fixed: the coin is fair, so each flip has 2 equally likely sides.

    The probability of heads comes from the coin itself, not from earlier results.

  2. Ask whether the four heads changed the coin: they did not, because a coin has no memory.

    Independent random flips cannot even out by changing the next probability.

  3. Write the probability: P(H)=12.

    One side out of two equally likely sides is heads.

  4. Size the run: four heads in a row has probability (12)4=116.

    That is unusual for any particular four flips but turns up regularly over a long session, so it is no evidence of a trick coin.

Answer: 12

Takeaway: A fair coin has no memory. After any run, the next flip is still heads with probability 12.

Example 2 — A predictable tilt

Standard

A coin lands heads 90 times in 100 flips. Is that just randomness?
Show the steps and the answer
  1. Find what a fair coin would give: about 100×12=50 heads.

    Randomness predicts the long-run share, so compare the result with that share.

  2. Find the gap: 90−50=40 extra heads.

    In 100 fair flips the count usually lands between about 40 and 60, so 40 extra heads is far outside ordinary wobble.

  3. Conclude: a lean this strong is a pattern, not random variation.

    Betting on heads with this coin would win about 9 times in 10, which makes it predictable, the opposite of random.

Answer: No — that points to a biased coin

Takeaway: Random results wander near the expected share. A result far outside that range points to bias.

Example 3 — Find the mistake: which record is made up?

Application

One student flipped a real coin 20 times; another invented 20 flips. Record A: HTHTHHTHTTHTHTHHTHTH. Record B: HHTHTTTTHHTHHHHTTHTT. Tess says: "B is the made-up one, because a real coin wouldn't land the same way four times in a row." Find the mistake and correct it.
Show the steps and the answer
  1. Find the longest runs: A never has more than 2 in a row; B has TTTT and HHHH.

    Tess is judging by runs, so check what runs real coins produce.

  2. Use what fair coins do: in 20 fair flips, the chance of at least one run of 4 or more is about 77%.

    Runs of 4 are normal in real flips, so B's runs are not evidence of invention.

  3. Count switches between H and T: A switches 16 times out of 19 chances; B switches 9 times.

    A fair coin switches about half the time, so roughly 9 or 10 switches is expected.

  4. Size A's pattern: a fair coin switches 16 or more times in 19 with probability 1160219≈0.002.

    Too much alternation is the tell-tale sign of a person trying to look random.

Answer: Record A is the made-up one: it alternates far too often, while B's runs of four are typical of a real coin.

Takeaway: Real randomness produces long runs. A sequence that avoids them is the one that looks invented.

Where You'll See It

  • Shuffling a deck of cards

    Mathematicians Dave Bayer and Persi Diaconis showed that about seven riffle shuffles are needed to make a 52-card deck close to random. With fewer, traces of the old order survive and a careful player could exploit them, which is why casinos use many shuffles or shuffling machines.

  • Choosing students for a survey

    Asking only your friends tilts the answers toward people like you. Drawing 50 names at random from a list of all 800 students gives every student the same chance, 50800=116, of being asked, so no group is favoured.

  • Lottery numbers that seem overdue

    Lottery balls have no memory, so a number that has not come up in months is no more due than any other. In a 6-from-49 lottery, every set of six numbers has the same 1 in 13,983,816 chance, including 1, 2, 3, 4, 5, 6.

Common Mistakes

  1. The slip: Expecting random sequences to alternate neatly

    Instead: real randomness produces streaks and clusters.

  2. The slip: Believing past outcomes affect the next independent one

    Instead: randomness has no memory.

  3. The slip: Confusing randomness (no pattern) with bias (a consistent tilt)

    Instead: bias is predictable, randomness is not.


Mini Practice

Try these on your own. Tap Reveal when you want to check.

  1. A fair die is rolled 600 times. About how many 4s should you expect? Can you predict what roll number 601 will show?

    Hint: Long-run rates are predictable; single rolls are not.

  2. A spinner is one quarter red. You spin it 40 times. How many reds do you expect, and would 13 reds be surprising?

    Hint: Find the long-run expectation, then think about normal random variation.

  3. Which sequence of 10 fair coin flips is more likely: HTHTHTHTHT or HHTHTTTHHT?

    Hint: Find the probability of any one exact sequence of 10 flips.

  4. In 20 fair coin flips, is it unusual to see 4 of the same side in a row somewhere?

    Hint: Random sequences are not as neat as people expect.

  5. Spinner A is half red and lands on red 52 times in 100 spins. Spinner B is also half red but lands on red 80 times in 100. Which looks like randomness, and which looks like bias?

    Hint: Compare each result with the expected 50.

  6. Nia sees the flips HHHHTTHHH and says: "This can't be random, because a real coin would switch back and forth more." Correct her.

    Hint: Do random sequences have to alternate?

Want the full set?

50 practice questions for this concept — free to try, every one with a complete worked solution showing the why, not just the answer.

Frequently Asked Questions

Can a computer really pick a random number?

Most computer random numbers are pseudo-random: a formula produces a long string of numbers that pass every test for randomness, but the string would repeat exactly if you started the formula from the same seed value. For games and simulations that is fine. For passwords and online banking, computers mix in physical noise, such as tiny timing differences in the hardware, so the result cannot be predicted.

If I try to write a random list of heads and tails, why does it come out wrong?

People switch sides too often. Asked to write 100 random flips, most people avoid runs longer than three, because a long run feels planned. Yet in 100 real flips, a run of at least six of the same side happens about 4 times in 5. Teachers use this as a trick: they can often tell a made-up list from real flips just by finding the longest run.

Is shuffling a deck seven times really enough?

For the usual riffle shuffle, where you split the deck and let the halves fall together, mathematicians Dave Bayer and Persi Diaconis showed in 1992 that about seven shuffles mix a 52-card deck well. With fewer, traces of the old order remain, which a skilled card player could use. Overhand shuffling, sliding small clumps from hand to hand, is much weaker and needs well over a thousand shuffles to do the same job.

Learning Path

← Before

Probability
Randomness

You are here

You need Probability first: a number from 0 to 1 for how likely an outcome is. Randomness explains what that number looks like in action: single results are unpredictable, long-run shares are not. Next, Sample Space lists every outcome a random process can produce, and the Law of Large Numbers (Intuition) explains why the long-run shares settle down.

See Also