T-Distribution in AP Statistics

In AP Statistics, the t-distribution is a symmetric, bell-shaped distribution with heavier tails than the standard normal. You use it for confidence intervals and significance tests about population means when the population standard deviation is unknown and you estimate it with the sample standard deviation s.

What is the t-Distribution?

The t-distribution is the curve you use for inference about means. It looks like the standard normal curve. It's centered at 0, symmetric, and bell-shaped, but it has heavier tails. The extra weight in the tails is the price you pay for not knowing σ. In real life you almost never know the population standard deviation, so you plug in the sample standard deviation s. That estimate adds uncertainty, and the t-distribution builds that uncertainty right into the curve.

There isn't just one t-distribution. There's a whole family, and each member is set by its degrees of freedom (df). For a one-sample t procedure (or matched pairs), df = n − 1. Small df means fatter tails and bigger critical values. As df grows, s becomes a more reliable estimate of σ, and the t-distribution gets closer and closer to the standard normal. A useful way to picture it is that the t-distribution is the normal curve with a built-in "we're not totally sure" penalty, and that penalty shrinks as your sample grows.

Why the t-Distribution matters in AP® Statistics

The t-distribution is the engine behind every means procedure in the inference-for-means topics: constructing a confidence interval for a population mean or mean difference (Topic 4.2), carrying out a test for a population mean or mean difference (Topic 4.5), constructing a confidence interval for the difference between two population means (Topic 4.7), and carrying out a test for the difference between two population means (Topic 4.10). In all four, you find a t* critical value for intervals or a t test statistic and p-value for tests, using the right degrees of freedom. If you mix up t and z, or use the wrong df, your interval width and p-value come out wrong. That's why this one idea quietly decides points across a big chunk of the inference questions on the exam.

How the t-Distribution connects across the course

Critical Value t* (Unit 4)

For a confidence interval, t* is the cutoff that captures the middle C% of the t-distribution with your df. Because the tails are heavier, t* is always a bit bigger than the matching z*. That makes t-intervals slightly wider. With df = 9, the 95% t* is about 2.262, compared to z* = 1.96.

Standard Error (Unit 4)

The standard error s/√n uses s instead of σ, and that swap is the whole reason you need t instead of z. Any time your formula has s in the denominator for a mean, the t-distribution is the right reference curve. For two samples, the standard error is √(s₁²/n₁ + s₂²/n₂), and you still use t.

Matched Pairs Design and Dependent Samples (Unit 4)

Paired data looks like a two-sample problem, but it's really a one-sample t procedure on the differences. You compute each pair's difference, then run a one-sample t with df = (number of pairs) − 1. Spotting dependence is what tells you which version of t to use.

Skewness and the Normal/Large Sample Condition (Unit 4)

The t-distribution only works well if the sampling distribution of x̄ is roughly normal. If n is at least 30, you're fine. With a small sample, you need to check a graph of the data for strong skewness or outliers. Heavier tails in t account for estimating σ, not for messy, lopsided data.

Is the t-Distribution on the AP® Statistics exam?

On multiple choice, the t-distribution shows up in a few predictable ways. You'll identify the correct procedure (two-sample t vs. paired t vs. z), pick the right degrees of freedom, compare t* to z*, or spot a FALSE statement about conditions. A typical stem gives two independent random samples with their own n, x̄, and s, like reaction times under two conditions (n = 15 in one group) or two study methods (n₁ = 28, n₂ = 32). Then it asks you to build an interval or interpret a result. Another common stem asks under what condition a two-sample t interval is appropriate. The answer hinges on random sampling, independence, and either large samples or approximately normal populations.

On free response, you have to name the procedure, check conditions (Random, 10% condition, Normal/Large Sample), state df, calculate the t statistic or interval, and interpret in context. The 2026 FRQ Q6, set in a baseball context about the relationship between hitting and getting on base, used the term. That's a reminder that t-based reasoning can appear inside a longer investigative-task question, not only in a plain means problem.

The t-Distribution vs Standard Normal (z) Distribution

Both curves are symmetric and bell-shaped and centered at 0, but they're used in different situations. Use z for proportions, and for means only in the rare case where σ is known. Use t for means when you estimate σ with s, which is basically always. The t-distribution has heavier tails and depends on degrees of freedom. The z-distribution is one fixed curve. As df gets large, t approaches z, so the difference matters most with small samples.

Key things to remember about the t-Distribution

• Use the t-distribution for confidence intervals and tests about means whenever you estimate the population standard deviation with the sample standard deviation s.

• The t-distribution is symmetric and bell-shaped like the normal curve, but its heavier tails reflect the extra uncertainty from estimating σ.

• Degrees of freedom determine which t-distribution you use, and for a one-sample or matched pairs procedure df equals n − 1.

• As degrees of freedom increase, the t-distribution gets closer to the standard normal, so t* shrinks toward z*.

• Matched pairs data uses a one-sample t procedure on the differences, while two independent samples use a two-sample t procedure.

• Before using t, check that the sample is random, that independence holds, and that either n is at least 30 or the data show no strong skewness or outliers.

Frequently asked questions about the t-Distribution

What is the t-distribution in AP Stats?

It's the symmetric, bell-shaped distribution with heavier tails than the normal that you use for inference about means when σ is unknown. Its exact shape depends on degrees of freedom, which equal n − 1 for a one-sample t procedure.

Is the t-distribution the same as the normal distribution?

No. They look similar, but the t-distribution has heavier tails and changes shape with degrees of freedom. It only becomes nearly identical to the standard normal when df is large.

When do you use t instead of z on the AP Stats exam?

Use t for means when you're using the sample standard deviation s, which covers basically every means problem on the exam. Use z for proportions. A z procedure for a mean only applies in the unusual case where the population standard deviation σ is given as known.

Do I need a large sample to use the t-distribution?

Not necessarily. If n is at least 30, the Normal/Large Sample condition is met. With a smaller sample like n = 15, you can still use t as long as a graph of the data shows no strong skewness or outliers.

How do you find degrees of freedom for a two-sample t test?

Calculators compute df with a formula that usually gives a non-integer value. A conservative option is to use the smaller of n₁ − 1 and n₂ − 1. For samples of 28 and 32, that gives df = 27. For matched pairs, use the number of pairs minus 1 instead.