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

Statistics · Grade 9-12

Normal Distribution

The normal distribution (bell curve) is a symmetric, bell-shaped probability distribution where most data clusters around the mean, with probabilities decreasing symmetrically toward the tails.


Quick Answer

The normal distribution (bell curve) is a symmetric, bell-shaped probability distribution where most data clusters around the mean, with probabilities decreasing symmetrically toward the tails. Two numbers fix the whole curve: the mean μ sets where the peak sits and the standard deviation σ sets how wide it spreads, written N(μ,σ2). Use it to model measurements shaped by many small random influences, such as heights, fill weights or repeated lab readings, and to turn a range of values into a probability. It is a model to check against the data, not a law every data set obeys.

Why This Matters

Once data are close to normal, the mean and standard deviation alone give the share of values in any range: a factory can predict what fraction of bolts will be too long, and a test maker can say which score marks the top 5%. Z-tables, confidence intervals and many tests later in statistics rest on this curve, and using it on skewed data such as incomes gives badly wrong tail probabilities.

Intuitive Explanation

Drop 1,000 marbles through a board of pegs. At each peg a marble bounces left or right at random. To end up far to the right it must bounce right almost every time, which is rare, so most marbles collect a mix of lefts and rights and land near the middle. The pile at the bottom is a bell: tall in the center, thinning evenly toward both edges. Any quantity built from many small, independent pushes, like adult height (many genes, diet, sleep) or the error in a measurement, piles up the same way.

Two numbers describe the whole bell. The mean μ is the center and the peak. The standard deviation σ is the distance from the center to where the curve changes from bending down to bending up; a small σ gives a tall, narrow bell and a large σ a low, wide one. The total area under the curve is always 1, meaning 100% of the data, and the share of values in a range is the area over that range. That is also why, for a continuous normal variable, the chance of one exact value such as 170.000 cm is 0.

Not every hump is normal. Waiting times at a clinic might peak near 10 minutes but stretch out to an hour on the right: one peak, but lopsided, so a normal model would underestimate long waits and could even predict negative ones. Before using the curve, check that a histogram is roughly symmetric with one peak and no long tail.

Think of it like this. A steady archer's misses behave like a normal curve: most arrows land near the aim point, big misses are rare, and misses to the left are as common as misses to the right. Where the archer aims is the mean; how steady the hands are is the standard deviation. Unlike real arrows, the model allows a tiny chance of a miss any distance away.

Core idea. When many small, independent effects add up, values pile into a symmetric bell fixed by its mean and standard deviation, and the area under that bell over a range is the probability of landing in it.


When to Use

Use the normal model when data are described as bell-shaped or approximately normal, or come from measurements with many small random influences, and the question asks for the probability or proportion in a range. First check that a histogram is roughly symmetric with one peak. For lopsided data such as incomes or wait times, describe the skew and use the data's own percentiles; for a quick estimate at exactly 1, 2 or 3 standard deviations, the Empirical Rule is enough.

Tip. Ask: Are you modelling a bell-shaped variable fixed by its mean and standard deviation, and finding the probability of a range as an area?

How to Recognize It

Before using the bell curve, check that the variable is plausibly bell-shaped and that the question wants a share or probability for a range of values.

  1. Is the variable a measurement shaped by many small, independent influences, such as heights or fill weights?

    Yes makes a normal model plausible. Quantities with a long tail, such as incomes or waiting times, are better described through their Skewness than forced into a bell.

  2. Does the question want the share of values, or the probability, between, above or below given values?

    Yes is an area under the normal curve. If it only asks how many standard deviations a value is from the mean, that is a Z-Score (Standard Score).

  3. Do the boundaries fall at awkward values, such as 163 cm, rather than exactly 1, 2 or 3 SDs from the mean?

    Yes needs the full normal curve, through a table or a calculator. Boundaries at whole numbers of SDs can be handled in your head with the Empirical Rule.

  4. Are the mean and standard deviation known, or can they be estimated from the data?

    Yes is required, because those two numbers fix the whole curve. Describing the outline of a histogram without any model is Distribution Shape.

Normal Distribution vs Empirical Rule vs Z-Score (Standard Score) vs Distribution Shape

These four are used together, which is why they blur: the shape suggests the model, z-scores move values onto it, and the empirical rule is its shortcut. Ask which job you are doing at this step.

Normal Distribution

Meaning
The normal distribution is a symmetric bell-shaped model whose mean sets the center and whose standard deviation sets the width.
Key test
Probabilities are areas under one exact curve, for any boundaries.
Formula
f(x)=12π σexp⁡(−(x−μ)22σ2)
Example
If battery life is N(40,52) hours, the share lasting over 47 hours is about 8%.

Empirical Rule

Meaning
The empirical rule gives the shares 68%, 95% and 99.7% within 1, 2 and 3 SDs of the mean of a normal distribution.
Key test
Boundaries sit at whole numbers of SDs, so no table is needed.
Formula
P(μ−2σ<X<μ+2σ)≈0.95
Example
For N(40,52) battery lives, about 95% last between 30 and 50 hours.

Z-Score (Standard Score)

Meaning
A z-score converts a raw value into standard deviations from the mean so it can be placed on the standard normal curve.
Key test
Gives a position; an area or probability needs one more step.
Formula
z=x−μσ
Example
A 47-hour battery has z=47−405=1.4.

Distribution Shape

Meaning
Distribution shape is the outline of real data, which may or may not look like a bell.
Key test
Describes the data you have, before any model is chosen.
Example
A histogram of 500 battery tests with two humps suggests two battery types, not one normal curve.

Formula & Notation

How to read it: N(μ,σ2) denotes a normal distribution with mean μ and variance σ2. The standard normal distribution is N(0,1) with μ=0 and σ=1.

Worked Examples

Example 1 — Share of underweight bags

Easy

A machine fills bags of chips with weights that are normally distributed, N(150,22) grams. What proportion of bags weigh less than 147 g?
Show the steps and the answer
  1. Read the parameters: μ=150 g and σ=2 g (the 4 in 22 is the variance).

    N(μ,σ2) lists the variance second, so the standard deviation is its square root.

  2. Measure 147 in standard deviations: 147−1502=−1.5.

    Every normal curve has the same shape in standard-deviation units, so one table covers all of them.

  3. Read the area to the left of −1.5 on the standard normal curve: about 0.0668.

    The proportion below a value is the area under the curve to its left.

  4. Interpret: about 6.7% of bags, roughly 67 in every 1,000.

    A sketch agrees: 147 is a bit beyond one standard deviation below the mean, where the tail holds well under 16%.

Answer: About 0.067, or 6.7% of bags weigh less than 147 g.

Takeaway: For a normal variable, convert the cut-off to standard deviations from the mean, then the answer is an area under one standard curve.

Example 2 — A mean and SD do not make data normal

Standard

Household incomes in a city have mean $72,000 and standard deviation $40,000, and the histogram has a long right tail. A student uses a normal model to estimate that 16% of households earn under $32,000. Is the normal model appropriate here?
Show the steps and the answer
  1. Note where the student's 16% comes from: $32,000 is 32000−7200040000=−1 SD, and about 16% of a normal curve lies below −1.

    The calculation is correct for a normal curve; the question is whether the curve fits.

  2. Test the model at $0: 0−7200040000=−1.8, and the normal area below −1.8 is about 0.036.

    A model that fits should not give a real chance to an impossible value.

  3. Interpret: the model says about 3.6% of households earn less than nothing.

    Incomes cannot be negative; the long right tail means the data are skewed, not bell-shaped.

  4. Use the income data itself, for example its percentiles, to find the share under $32,000.

    A skewed distribution needs a method that does not assume symmetry.

Answer: No. The data are right-skewed, and the normal model would even put about 3.6% of households below $0, so its 16% estimate cannot be trusted.

Takeaway: Every data set has a mean and a standard deviation; only a roughly symmetric, single-peaked one earns the normal model.

Example 3 — Find the mistake: 95% for one SD

Application

Scores on a large exam are approximately normal with mean 70 and standard deviation 8. Jade writes: 'About 95% of students scored between 62 and 78.' Find the mistake and correct it.
Show the steps and the answer
  1. Measure the interval: 62=70−8 and 78=70+8, so it runs from −1 to +1 SD.

    The share of a normal curve depends only on how many standard deviations the interval spans.

  2. Recall the shares: about 68% within 1 SD, 95% within 2 SD, 99.7% within 3 SD.

    Jade attached the 2 SD share to a 1 SD interval.

  3. Correct both claims: 62 to 78 holds about 68%; the 95% interval is 70±16, from 54 to 86.

    Doubling the distance from the mean is what takes the share from 68% to 95%.

Answer: About 68% of students scored between 62 and 78; about 95% scored between 54 and 86.

Takeaway: Match the share to the number of standard deviations: 1 SD is 68%, 2 SD is 95%.

Where You'll See It

  • Class height chart

    If you measure the height of every kid in 4th grade and make a histogram, most kids are near the average height and only a few are very short or very tall — the bars form a bell shape.

  • Standardized test scores

    On a big test like the SAT, scores are built to follow a bell curve: lots of students score near the middle and fewer score at the very top or very bottom, so a single mean plus SD describes the whole pattern.

  • Bag of chips weights

    A factory aims for 150 g per bag, but tiny errors make each bag a little over or under. The weights cluster around 150 g and thin out toward the extremes, forming a normal distribution.

Common Mistakes

  1. The slip: Any data that has a 'hump in the middle' is a normal distribution.

    Why it's tempting: Many curves rise in the middle and fall on the sides, so a single bump looks like proof. But a bell shape must also be symmetric and have the exact tail behavior; skewed or heavy-tailed humps are not normal.

    Instead: Only call data normal if it is symmetric about the mean AND its spread matches the bell shape (e.g., roughly 68% within one SD). Check with a histogram or normal-probability plot, not just 'it has a peak.'

  2. The slip: The normal distribution tells you the most common single value, like the mode of a list.

    Why it's tempting: The peak of the curve is at the mean, which feels like 'the answer.' But a continuous distribution describes probabilities over ranges, not the chance of one exact value.

    Instead: Use the normal distribution to find the proportion of data in an interval (between two values), not the probability of one precise value — for a continuous variable, P(X = exact value) is essentially 0.

  3. The slip: If the mean and SD are given, the data must be normal.

    Why it's tempting: Every data set has a mean and an SD, so it feels like those two numbers define a bell curve. But mean and SD exist for skewed and uniform data too.

    Instead: Mean and SD are just summaries; normality is an additional assumption about shape. Verify the shape before applying normal-curve rules like 68-95-99.7.


Mini Practice

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

  1. Adult heights in a group are normal with mean 170 cm and standard deviation 8 cm. What height is 2 standard deviations above the mean?

    Hint: Add two standard deviations to the mean.

  2. IQ scores are normal with mean 100 and standard deviation 15. About what percent of people score between 85 and 115?

    Hint: How many standard deviations from the mean are 85 and 115?

  3. Bags of chips have weights that are normal with mean 28.0 g and standard deviation 0.5 g. About what percent of bags weigh less than 27.0 g?

    Hint: How many standard deviations below the mean is 27.0 g? Then use the tail.

  4. Exam scores are normal with mean 500 and standard deviation 100. A table gives P(Z<1.5)≈0.933. About what percent of students score above 650?

    Hint: Find the z-score of 650, then use the part of the curve above it.

  5. In a city, household incomes have mean $85,000 and median $62,000. Is a normal model a good fit?

    Hint: Where are the mean and median in a normal distribution?

  6. Clinic waiting times have mean 10 min and standard deviation 8 min. A student assumes they are normal and concludes: "About 16% of patients wait less than 2 minutes." What is wrong with this reasoning?

    Hint: Under a normal model, what would mean minus 2 SD be?

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

Why do so many measurements come out roughly normal?

Because many of them are the sum of lots of small, independent effects, and sums like that tend toward a bell shape whatever the separate effects look like. One die gives a flat distribution, but the total of ten dice already looks like a bell, with totals near 35 common and totals near 10 or 60 very rare. This idea, made precise, is the central limit theorem, and it is why sample means are so often treated as normal.

Can a normal model give a negative height or a negative waiting time?

In principle yes, because the curve never quite reaches zero on either side. For heights with mean 170 cm and SD 10 cm, a negative height is 17 SDs below the mean, so the chance is too small to matter. For waiting times with mean 3 minutes and SD 3 minutes, though, the model would put about 16% of waits below zero, which is impossible. That is a sign the normal model is wrong there.

Is "grading on a curve" the same as the normal curve?

Not really. Grading on a curve usually means handing out grades by rank, for example the top 10% get an A, whatever the actual scores are. Some teachers set those shares to match a normal curve, but the class scores themselves may be lopsided or have two clusters. The normal distribution is a model to check against data, while grading on a curve forces a fixed pattern of grades onto any data.

Learning Path

Normal Distribution

You are here

Start from Distribution Shape (symmetric versus skewed, one peak or several) and Standard Deviation (the typical distance from the mean). The normal distribution adds a precise curve that turns those two numbers into areas. Next, Z-Scores convert any normal variable to the standard normal N(0,1), so one table serves every problem, and the Empirical Rule gives the 68-95-99.7 shortcut.

See Also