Section II is four free-response questions worth 10 points each: 90 minutes, 50 percent of the exam score, so each question carries 12.5 percent on its own. This guide covers the redesigned course (2026 CED, first exam May 2027). The retired 9-unit course ran six shorter FRQs with an investigative task, so most older prep materials describe a section that no longer exists. A graphing calculator with statistical capabilities is expected in both sections, and the formula sheet and tables are provided in both. The points sit in statistical reasoning stated in context, not in arithmetic.
Questions
4
10 points each
Time
90 min
about 22 per question
Weighting
50%
12.5% per question
Provided
Sheet + tables
calculator expected, both sections
The four questions
What each question asks
1
Multi-Focus: Design and Data Collection
Practices 1 and 2 · Formulate Questions · Collect Data
Study design from the ground up: writing and recognizing statistical questions, sampling methods, sources of bias, and what the design permits you to conclude. Expect to justify why a design choice matters, not just name it.
Random selection vs random assignment: selection lets you generalize to the population; assignment lets you talk about cause and effect. They are different powers earned by different acts.
Name the mechanism of a bias and its likely direction in this study: "nonresponse from the busiest commuters likely understates the mean commute time", not just "nonresponse bias".
Scope of inference: an observational study cannot support a causal claim, no matter how large the sample.
2
Multi-Focus: Analysis and Interpretation
Practices 3 and 4 · Analyze Data · Interpret Results
Distributions, probability, and random variables, with every computed value read back into the situation. The calculator does the arithmetic; the points are in the sentences.
Describe distributions with shape, center, variability and unusual features, in context with units, and use comparative language when two groups are shown.
Define the random variable and state its distribution before computing a probability, so the work is checkable.
A number is not an answer. An interpretation names the variable, the units, and the group it describes.
3
Inference
A complete hypothesis test or confidence interval
A full inference procedure from start to finish: hypotheses or parameter definition, the procedure by name, conditions verified, mechanics, and a conclusion in context. This is the most predictable question on the exam, which makes it the most trainable.
Name the procedure: "one-sample t-interval for a mean" or "two-proportion z-test", so the reader knows what you are doing before you do it.
Verify conditions with numbers: np0 = 45 ≥ 10 and n(1 - p0) = 55 ≥ 10, not "large counts: yes".
Link the p-value to the decision: "because p = 0.03 < α = 0.05, we reject H0", then state what that means about the parameter in context.
4
Multi-Focus: Integrated
Practices 2, 3, and 4 · Design to conclusion
One scenario carried from data collection through analysis to interpretation. Later parts often lean on earlier ones, so read the whole question before writing anything.
Answer every part in the scenario's terms. The context-sentence discipline applies in each part, not just the last one.
Stay consistent with your own numbers: if an earlier part went wrong, a later conclusion that correctly follows from it can still earn its points.
Budget time: four questions in 90 minutes is about 22 minutes each. A 10-point question rewards finishing every part more than perfecting one.
Earning the points
What the rubric rewards
Context in every conclusion. Name the variable, the units, and the population or groups. "The mean is 45" earns less than "the mean commute time for this district's teachers is 45 minutes".
Link the p-value to the decision explicitly. "Because p = 0.03 < α = 0.05, we reject H0; there is convincing evidence that..." is the full sentence the rubric looks for.
Name the inference procedure before the mechanics, and define the parameter in words. An unnamed procedure forces the reader to reverse-engineer your intent.
Verify conditions with numbers. Random: stated in the stem. Ten percent: n = 100 is less than 10 percent of the district. Large counts: np0 = 45 ≥ 10. Conditions named without numbers read as unverified.
Let the calculator compute; you interpret. A graphing calculator with statistical capabilities is expected in both sections, and the formula sheet and tables are provided in both. Time saved on arithmetic is time for sentences.
Conclude about parameters, not samples. Inference concludes about the population mean or proportion; the sample statistic was never in question.
The fastest way to stop losing FRQ points
Most free response points are lost to a small set of documented misconceptions: a p-value read as the probability the null is true, a confidence level pinned to one interval, a causal claim from an observational study, a conclusion with no context. The lessons target exactly these, one misconception at a time.
College Board publishes free response questions with scoring guidelines and sample student responses on the AP Statistics exam page. One caution for the class of 2027 and later: questions released before 2027 follow the retired 9-unit course, which ran six FRQs including an investigative task. The inference-writing habits they train transfer directly; the section structure and point values do not. The Course and Exam Description carries the redesigned format.
Common questions
FAQ
How many free response questions are on AP Statistics?
Four, each worth 10 points, in a 90-minute section worth 50 percent of the exam score, so each question is 12.5 percent of the total. This is the redesigned exam, first given May 2027; the retired course had six FRQs including an investigative task.
Can I use a calculator on the AP Statistics FRQs?
Yes. A graphing calculator with statistical capabilities is expected in both sections of the exam, and the formula sheet and tables are provided in both sections as well. There is no no-calculator part.
What is Question 3 always about?
Inference: a complete hypothesis test or confidence interval. Name the procedure, define the parameter, verify the conditions with numbers, run the mechanics, and conclude in context with the p-value linked to the decision.
Do older released AP Statistics FRQs still help?
Yes, with a caution: FRQs released before 2027 follow the retired 9-unit course, with six questions and different point values. The reasoning and writing they train still score points; the format they model is gone.
Do I lose points for a conclusion without context?
Routinely. A conclusion that never names the variable, units, or population reads as generic and earns partial credit at best. Context is the most affordable point on the section.
