Review AP Statistics Unit 4 to build your full toolkit for inference about population means, from sampling distributions and the t-distribution through confidence intervals and significance tests for one mean, matched pairs, and two independent groups. This unit is the quantitative counterpart to Unit 3 and forms the backbone of the inference portion of the AP exam.
Use the topic guides, key terms, and practice questions available on Fiveable to work through each procedure in order before tackling two-sample inference.
Unit 4 is where AP Statistics moves from proportions to means. Because the population standard deviation is almost never known for quantitative variables, every procedure here uses the t-distribution rather than the standard normal. The unit is organized in two parallel tracks: inference for one population mean (including matched pairs) in Topics 4.1-4.5, and inference for the difference between two population means in Topics 4.6-4.10.
When you use the sample standard deviation s in place of the unknown population standard deviation sigma, you introduce extra variability. The t-distribution accounts for this by having heavier tails than the standard normal. As degrees of freedom (df = n - 1 for one sample) increase, the t-distribution approaches the standard normal.
One-sample t-procedures apply when you have a single quantitative sample. Matched pairs designs produce one sample of differences (subtract within each pair), so you still run a one-sample t-procedure on those differences. Two independent groups require two-sample t-procedures with a pooled standard error formula.
Before any t-interval or t-test, verify: (1) Randomization: data come from a random sample or randomized experiment. (2) 10% condition: n is at most 10% of the population (for observational data). (3) Normal/large sample: population is approximately normal, or n is at least 30, or n is less than 30 with no strong skew or outliers in the sample.
Every confidence interval and significance test in Unit 4 uses sample data to draw conclusions about a population parameter, either mu, mu_d, or mu_1 minus mu_2. The p-value is always computed assuming the null hypothesis is true, and conclusions must be stated in context using non-definitive language that references the population, not just the sample.
Calculate mu_x-bar = mu and sigma_x-bar = sigma / sqrt(n), verify conditions, and apply the Central Limit Theorem to determine the shape of the sampling distribution.
Build a one-sample t-interval using x-bar plus or minus t*(s / sqrt(n)) with df = n - 1; apply the same procedure to matched-pairs differences.
Interpret a t-interval in context, identify the relationships among confidence level, sample size, and margin of error, and use the interval to support or refute a claim.
State H_0 and H_a in terms of mu or mu_d, identify the one-sample t-test or matched-pairs t-test, and verify the three conditions before calculating anything.
Calculate t = (x-bar - mu_0) / (s / sqrt(n)) with df = n - 1, find the p-value, compare to alpha, and write a conclusion in context referencing the population.
Find the mean and standard deviation of the sampling distribution of x-bar_1 minus x-bar_2 using the addition rule for variances, and verify normality conditions for each group.
Apply the two-sample t-interval formula (x-bar_1 - x-bar_2) plus or minus t* times sqrt(s_1^2/n_1 + s_2^2/n_2), with df from technology, to estimate mu_1 - mu_2.
Interpret the two-sample t-interval in context; determine whether 0 is inside or outside the interval to decide if there is evidence of a difference between the two population means.
State H_0: mu_1 = mu_2 and the appropriate H_a, identify the two-sample t-test, and verify randomization, 10%, and normal/large sample conditions for both groups.
Calculate t = (x-bar_1 - x-bar_2) / sqrt(s_1^2/n_1 + s_2^2/n_2) with df from technology, find the p-value, and state a conclusion in context referencing both populations.
Open this guide for a closer review of the topic.
The sampling distribution of x-bar describes how sample means vary across repeated random samples. Its center equals the population mean and its spread shrinks as sample size grows. Knowing its shape lets you calculate probabilities about x-bar before doing formal inference.
When sigma is unknown, estimate mu with a one-sample t-interval. For matched pairs, compute differences within each pair first, then treat those differences as a single sample. Interpret the interval in context and use it to justify or refute a claim about the population mean.
A one-sample t-test evaluates whether sample data provide convincing evidence against a null hypothesis about mu. For matched pairs, the test runs on the sample of differences. After calculating the test statistic and p-value, compare the p-value to alpha and state a conclusion in context.
When comparing two independent groups, the relevant statistic is x-bar_1 minus x-bar_2. Its sampling distribution has a predictable center and spread, and it is approximately normal under the same conditions used for one-sample means applied to each group.
Use a two-sample t-interval when you have two independent samples and want to estimate mu_1 minus mu_2. The point estimate is x-bar_1 minus x-bar_2, and the margin of error uses a pooled standard error. Whether 0 is inside or outside the interval determines whether you have evidence of a difference.
A two-sample t-test assesses whether two independent groups have different population means. The null hypothesis is that the two means are equal (difference equals 0). The test statistic divides the observed difference in sample means by the standard error of that difference.
Selecting the correct procedure is a required skill on the AP exam. The key distinctions are: how many groups, whether the groups are paired or independent, and whether you are estimating (interval) or testing (significance test).
| Scenario | Procedure | Parameter | df |
|---|---|---|---|
| One sample | One-sample t-interval or t-test | mu | n - 1 |
| Matched pairs | One-sample t-interval or t-test on differences | mu_d | n - 1 |
| Two independent samples (interval) | Two-sample t-interval | mu_1 - mu_2 | Technology |
| Two independent samples (test) | Two-sample t-test | mu_1 - mu_2 | Technology |
| Term | Definition |
|---|---|
| t-Distribution | A family of symmetric, bell-shaped distributions with heavier tails than the standard normal, used when sigma is unknown. Identified by degrees of freedom; approaches the standard normal as df increases. |
| One Sample T-test | A significance test that uses x-bar and s to test a claim about one population mean mu when sigma is unknown; test statistic is t = (x-bar - mu_0) / (s / sqrt(n)) with df = n - 1. |
| Two-sample t-test | A significance test that uses two independent samples to test whether mu_1 - mu_2 = 0; test statistic is t = (x-bar_1 - x-bar_2) / sqrt(s_1^2/n_1 + s_2^2/n_2) with df from technology. |
| Matched Pairs Design | A design in which two measurements are taken on the same subject or on deliberately paired subjects; inference is performed on the sample of within-pair differences using one-sample t-procedures. |
| dependent samples | Samples where observations are paired or related, such as before-and-after measurements on the same subjects; require matched pairs analysis rather than two-sample procedures. |
| Independent groups | Two groups whose observations are unrelated to each other; required for two-sample t-procedures. Contrast with matched pairs, where observations are linked. |
| 10% Condition | Requires that the sample size is no more than 10% of the population when sampling without replacement, supporting approximate independence between observations. |
| randomness condition | Requires that data come from a random sample or randomized experiment before any t-interval or t-test is applied. |
| Critical Value | The t* multiplier determined by the confidence level and degrees of freedom; used to compute the margin of error in a t-interval. |
| Confidence Level | The long-run percentage of t-intervals from repeated random samples that would capture the true population mean or mean difference. |
| Significance Level | The threshold alpha used to decide whether to reject H_0; if the p-value is less than or equal to alpha, the result is statistically significant. |
| Statistically Significant | A result is statistically significant when its p-value is less than or equal to alpha, providing evidence against the null hypothesis about mu or mu_1 - mu_2. |
| One-tailed test | A hypothesis test with a directional alternative hypothesis (H_a: mu < mu_0 or H_a: mu > mu_0); the p-value is the area in one tail of the t-distribution. |
| Independence Condition | Requires that observations within a sample are independent of each other, supported by the 10% condition for random samples or by random assignment in experiments. |
If the same subjects are measured twice or subjects are deliberately paired, use a one-sample t-procedure on the differences. Running a two-sample t-test on paired data ignores the pairing and inflates variability, leading to incorrect results.
The confidence level describes the long-run capture rate across repeated samples, not the probability that this specific interval contains the parameter. Avoid saying 'there is a 95% probability that mu is in this interval' after the interval is computed.
For matched pairs and two-sample procedures, the sign of the interval or test statistic depends on which group is subtracted from which. Always define the order of subtraction when stating the parameter and interpret results accordingly.
On free-response questions, all three conditions (random, 10%, normal/large sample) must be checked explicitly with reference to the context. Simply listing condition names without connecting them to the problem earns no credit.
Because sigma is unknown for quantitative variables, t-procedures are required in Unit 4. Using z* as the critical value or the standard normal to find p-values is incorrect unless the problem explicitly states that sigma is known.
AP Statistics free-response questions on inference for means typically require all four steps: state the parameter and procedure, verify conditions with context-specific justification, calculate the test statistic or interval, and write a conclusion in context. Partial credit is awarded at each step, so skipping condition checks or writing a conclusion without referencing the population costs points.
Multiple-choice and free-response questions frequently present a scenario and ask which t-procedure is appropriate. You must distinguish between one-sample, matched pairs, and two-sample situations based on whether observations are paired or independent, and justify your choice by referencing the study design.
The AP exam consistently tests whether students can write correct, context-specific interpretations of confidence intervals and p-values. For intervals, the interpretation must reference the population parameter and use 'confident' language. For p-values, the interpretation must state the probability is computed assuming H_0 is true and describe the event in terms of the variable and population.
Open the individual guides for Unit 4 when you want a closer review of one topic.
browse guidesUse AP-style practice after you review the notes so you can check what you understand.
start practicePractice free-response reasoning and compare your answer with scoring guidance.
practice FRQsWatch past review streams filtered to Unit 4 when you want a video walkthrough.
open videosOpen the Fiveable one-page unit review, then explore visual cheatsheets for a quick refresher.
open unit cheatsheetEstimate your broader AP score goal after you review the course and exam format.
open calculatorAP Stats Unit 4, Inference for Quantitative Data: Means, contains 10 topics covering sampling distributions for sample means; one-sample and matched-pairs t intervals and tests; two-sample t intervals and tests; standard error; margin of error; p-values; and conclusions about population means or mean differences. These are the Fall 2026 CED topics used for the May 2027 exam.
Unit 4 accounts for 10–20% of the multiple-choice section. That range is a section weight, not a promise that every practice form will use the same percentage, so review the full unit rather than trying to predict one exact question count.
By the end of Unit 4, you should be able to decide whether a one-sample, matched-pairs, or two-sample t procedure applies, identify the parameter, verify randomization and distribution conditions, use the correct standard error, and interpret results in context. The exam rewards the method and the interpretation, so practice writing what a result means in the problem's context instead of stopping at a calculator output.
Common Unit 4 mistakes include treating paired data as independent samples, using a z procedure when the population standard deviation is unknown, failing to check for strong skew or outliers in small samples, or interpreting a confidence interval as a probability about a fixed parameter. Slow down long enough to identify the variables, population, and requested conclusion before calculating.
Start with the vocabulary and conditions, then mix short calculations with full-sentence interpretations. For each missed question, label whether the problem was choosing a method, doing the calculation, or interpreting the result. Finish with timed mixed practice so you have to recognize the method without a topic label.
Unit 4 uses the same inference logic as Unit 3 but applies it to quantitative responses and t distributions rather than proportions and z procedures.