Skewness in AP Statistics

In AP Statistics, skewness describes a distribution that is not symmetric, with a longer tail on one side. Right-skewed data have a long tail to the right, left-skewed data a long tail to the left. Strong skew in a small sample can make t-procedures for means invalid.

What is Skewness?

Skewness is about shape. A distribution is skewed when one side stretches out into a long tail and the other side is bunched up. The direction of the skew is named after the tail, not the hump. A right-skewed (positively skewed) distribution has most values piled up on the low end and a few large values trailing off to the right. Think income or hospital stay lengths. A left-skewed (negatively skewed) distribution has the long tail on the low end, like scores on a very easy test.

Skew pulls the mean toward the tail, because the mean feels every extreme value. The median barely moves. In right-skewed data the mean is usually bigger than the median. In left-skewed data it's usually smaller. That's why AP Stats wants you to report the median and IQR for skewed data and save the mean and standard deviation for roughly symmetric data. On AP Stats you describe skewness in words from a graph (histogram, dotplot, box plot, stemplot). You don't calculate a skewness coefficient.

Why Skewness matters in AP® Statistics

Skewness shows up twice in the course. First, you use it whenever you describe a distribution, along with center, spread, and unusual features. Second, and this is where it gets tested hard, it controls whether you can use t-procedures. In Topic 4.2 (Constructing a Confidence Interval for a Population Mean or Population Mean Difference), Topic 4.4 (Setting Up a Test for a Population Mean or Population Mean Difference), and Topic 4.9 (Setting Up a Test for the Difference Between Two Population Means), the Normal/Large Sample condition asks a simple question. Is the sampling distribution of the mean approximately normal? If n ≥ 30, the Central Limit Theorem handles it. If n < 30, you have to look at the sample data and confirm there's no strong skewness and no outliers. So skewness isn't just a vocabulary word. It's the thing that can sink an entire inference procedure.

How Skewness connects across the course

Normal/Large Sample condition for t-procedures (Topics 4.2, 4.4)

This is the biggest link. A small sample from strongly skewed data means the sample mean's distribution may not be normal, so the t-interval or t-test isn't trustworthy. Once n hits 30 or more, the Central Limit Theorem smooths the skew out of the sampling distribution, even when the data themselves stay skewed.

Two-sample and paired t-tests (Topics 4.4, 4.9)

Where you check for skew depends on the design. With independent groups, you check each sample separately, and both need to pass. With dependent (paired) samples, you ignore the two original lists and check the distribution of the differences.

Box plots and the 1.5×IQR rule (Unit 1)

A box plot shows skew at a glance. The median sits off-center in the box and one whisker runs much longer than the other. The 1.5×IQR rule then flags the outliers that often live out in that long tail. Those outliers are exactly what the normality check warns you about.

Mean vs. median as measures of center (Unit 1)

Skew is a tug-of-war that only the mean loses. The long tail drags the mean toward it while the median stays put. That's why the mean-versus-median comparison is a quick clue about which way the data are skewed.

Is Skewness on the AP® Statistics exam?

Multiple choice. Expect condition-checking stems. One common setup describes recovery times with strong right-skewness in two groups of sizes 12 and 15, then asks which statement is correct. The key insight is that both samples are small and strongly skewed, so a two-sample t-test isn't appropriate. Another flips it around. Paired differences from 40 runners are fine because n ≥ 30 even if the differences aren't perfectly normal. You'll also see "Which scenario would NOT allow a valid t-test?" The answer is usually the small sample with strong skew or outliers.

Free response. Skewness appears in released FRQs in both of its roles. 2019 FRQ Q1 gave a histogram of the sizes of 20 residence-hall rooms, the classic "describe the distribution" setup where shape (including skew) earns credit. 2021 FRQ Q1 focused on hospital length of stay and unusually short or long stays, which connects skew to outliers. 2024 FRQ Q6 asked about estimating a mean whistle price, the kind of inference task where you have to justify the normality condition.

What you have to DO: name the direction of skew in context ("the distribution of room sizes is skewed right"). Explain how it affects the mean versus the median. For inference with n < 30, state that the sample shows no strong skewness or outliers, or explain that the condition fails because it does.

Skewness vs Right-skewed vs. left-skewed

This trips people up all the time. The direction of skew is named for where the tail points, not where the peak is. A right-skewed distribution has its hump on the LEFT and its long tail stretching RIGHT, toward larger values. A quick check is that the mean gets pulled toward the tail. Mean bigger than median suggests right skew, and mean smaller than median suggests left skew.

Key things to remember about Skewness

• Skewness means a distribution is lopsided, with a longer tail on one side than the other.

• The direction of skew is named for the side the tail stretches toward, so right-skewed data have a long tail of large values.

• Skew pulls the mean toward the tail while the median stays put, so use the median and IQR to summarize strongly skewed data.

• For t-procedures with n < 30, you must check that the sample data show no strong skewness or outliers before trusting the results.

• When n is 30 or more, the Central Limit Theorem makes the sampling distribution of the mean approximately normal even if the data are skewed.

• For two independent samples, check skewness in each group separately, and for paired data, check it in the distribution of differences.

Frequently asked questions about Skewness

What is skewness in AP Statistics?

Skewness describes a distribution that isn't symmetric because one side has a longer tail. You identify it from a graph, like a histogram or box plot, and describe it in context, for example "the length of hospital stays is skewed right."

Does skewed data mean I can't use a t-test?

No, not automatically. If the sample size is 30 or more, the Central Limit Theorem lets you proceed even with skewed data. The problem is a small sample with strong skew or outliers, like two strongly right-skewed groups of 12 and 15, where the normality condition fails.

How is right skew different from left skew?

Right-skewed data have a long tail toward larger values, and the mean is usually greater than the median. Left-skewed data have a long tail toward smaller values, and the mean is usually less than the median. The skew always points the way the tail goes.

How do you tell if a box plot is skewed?

Look for one whisker that's much longer than the other and a median that sits closer to one end of the box. If the right whisker is long and the median is near the left edge of the box, the distribution is skewed right, and the 1.5×IQR rule may flag outliers in that tail.

Is skewness on the AP Statistics exam?

Yes. You'll describe shape in "describe the distribution" questions, like the 2019 FRQ Q1 histogram of room sizes. You'll also judge it when checking conditions for t-intervals and t-tests on means, the kind of reasoning that inference questions like 2024 FRQ Q6 call for.