Normality in AP Statistics

In AP Statistics, normality is the condition that the sampling distribution of a statistic is approximately normal. You check it before using z or t procedures. For proportions, you verify it with the Large Counts condition (at least 10 expected successes and 10 expected failures).

What is Normality?

Normality is the idea that a distribution has that familiar bell shape. It is symmetric, it peaks at the mean, and its tails thin out evenly on both sides. On the AP exam, though, the word almost always shows up as a condition for inference. You aren't asking whether the raw data look like a bell. You're asking whether the sampling distribution of your statistic (like p̂ or p̂₁ − p̂₂) is approximately normal. That's what lets you use a z critical value or calculate a z-score and a p-value.

You can't see a sampling distribution directly, so you check normality with a rule of thumb. For proportions, that rule is the Large Counts condition. For a confidence interval (Topics 3.3 and 3.10), you need at least 10 successes and 10 failures in each sample, using the observed counts np̂ and n(1 − p̂). For a significance test (Topic 3.5), you plug in the hypothesized value p₀, so you need np₀ ≥ 10 and n(1 − p₀) ≥ 10. For means, normality comes from either a normal population or a large enough sample, which is where the Central Limit Theorem comes in.

Why Normality matters in AP® Statistics

Normality is one of the three conditions you check before nearly every inference procedure, along with random sampling and independence. It's at the core of Topic 3.3 (confidence interval for a population proportion), Topic 3.5 (setting up a test for a population proportion), and Topic 3.10 (confidence interval for the difference between two proportions). In each one, the formula only works if the sampling distribution is approximately normal. If that fails, the critical value and the p-value don't mean what you think they mean. Graders look for the condition to be named, checked with actual numbers, and connected to the conclusion. That makes normality one of the most reliable points you can earn, and one of the easiest to lose.

How Normality connects across the course

Large Counts and the Critical Value (Topics 3.3 & 3.10)

The z* you use in a confidence interval assumes a normal sampling distribution. Normality is the permission slip for that number. If the Large Counts check fails, z* is no longer the right multiplier, and your margin of error is off.

One Proportion Z Test (Topic 3.5)

In a test, you assume the null hypothesis is true, so you check normality with p₀ instead of p̂. It's the same idea with a different input, and mixing them up is a classic lost point.

Central Limit Theorem (inference for means)

The CLT is the reason normality works for means even when the population is skewed. With a large enough sample (n ≥ 30 is the usual guideline), the sampling distribution of x̄ is approximately normal. Released FRQs about means, like 2024 Q6 on whistle prices, depend on this logic.

Z-Score and Standard Deviation (normal distribution calculations)

When a variable is stated to be approximately normal, as with the melon diameters in 2017 FRQ Q3, you can turn values into z-scores and find probabilities. Normality is what makes that z-table or normalcdf answer trustworthy.

Is Normality on the AP® Statistics exam?

Multiple choice. Questions often ask which assumption is NOT required for a two-proportion confidence interval, or give you counts and ask whether conditions are met. For example, if 42 of 120 male students work part-time, you check that 42 and 78 are both at least 10. The wrong answer choices usually swap in a fake condition, like requiring the population itself to be normal, or confuse normality with the 10% condition.

Free response. Released FRQs such as 2017 Q5, 2019 Q4, and 2024 Q6 build inference into the task, and checking conditions is part of the scoring. Here's what earns credit:

• Name the condition, such as "Large Counts" or "normal sampling distribution."
• Show the actual numbers, like "180 and 70 are both ≥ 10."
• State what the check lets you conclude: the sampling distribution is approximately normal.
• For two samples, check all four counts.
• For means, cite n ≥ 30 or describe a sample graph with no strong skew or outliers.

If a problem tells you a variable is approximately normal (as 2017 Q3 does), use that to justify normal probability calculations.

Normality vs 10% Condition

Both conditions involve a number, so they blur together, but they protect different things. The Large Counts (normality) condition makes sure the sampling distribution is approximately normal so z methods work. The 10% condition (n ≤ 10% of the population) makes sure observations are close enough to independent when you sample without replacement. It belongs to the independence condition, not normality. A quick way to keep them apart: normality is about the shape of the sampling distribution, and the 10% condition is about the standard deviation formula being accurate.

Key things to remember about Normality

• On the AP exam, normality usually means the sampling distribution of the statistic is approximately normal, not that the raw data are bell-shaped.

• For proportions, you check normality with the Large Counts condition, which requires at least 10 expected successes and at least 10 expected failures.

• Use p̂ to check Large Counts for a confidence interval, but use the hypothesized p₀ for a significance test.

• For a difference of two proportions, all four counts (successes and failures in both samples) must be at least 10.

• For means, normality comes from a normal population, a sample of at least 30 thanks to the Central Limit Theorem, or a sample graph with no strong skew or outliers.

• The 10% condition supports independence, not normality, so don't use it as your normality check.

Frequently asked questions about Normality

What is the normality condition in AP Stats?

It's the requirement that the sampling distribution of your statistic is approximately normal before you use z or t procedures. For proportions, you verify it with the Large Counts condition, which needs at least 10 successes and 10 failures.

Does the population have to be normal to build a confidence interval for a proportion?

No. For proportions, you only need the sampling distribution of p̂ to be approximately normal, and the Large Counts condition takes care of that. A categorical variable doesn't even have a bell-shaped population to check.

Do you use p-hat or p0 to check Large Counts?

It depends on the procedure. For a confidence interval (Topics 3.3 and 3.10), use p̂ because you don't know the true p. For a one-proportion z test (Topic 3.5), use p₀ because you assume the null hypothesis is true.

How is the normality condition different from the 10% condition?

Normality (Large Counts) makes sure the sampling distribution is bell-shaped enough for z methods. The 10% condition checks that the sample is no more than 10% of the population, so observations are roughly independent. They are separate conditions, and the exam often tests whether you can tell them apart.

How do you check normality for a two-proportion confidence interval?

Check that successes and failures are each at least 10 in both samples. For example, with 180 of 250 households in County 1 and 210 of 300 in County 2, the counts 180, 70, 210, and 90 are all at least 10, so the sampling distribution of p̂₁ − p̂₂ is approximately normal.