AP Statistics is one of the most useful AP courses you will ever take, and one of the most teachable. The exam rewards clear reasoning, careful reading, and a small set of templates you can practice until they feel automatic. The 5 rate has hovered between 14 and 16 percent in recent years, and the curve gives partial credit on every free response question. With a focused 10 week plan, a clean calculator workflow, and a habit of writing in full sentences, a 5 is genuinely within reach. This guide walks through the entire course unit by unit, then gives you the templates, sample problems, and pacing strategy that move scores up two letter grades in a single semester.
Table of Contents
- Exam Format and Scoring in 2026
- 10 Week Study Timeline
- All Nine Units at a Glance
- Calculator Strategy: TI 84 vs Desmos
- FRQ Templates That Earn Full Credit
- Sample Practice Problems
- Test Day Strategy
- Common Mistakes to Avoid
- FAQ
Exam Format and Scoring in 2026
The AP Statistics exam runs three hours and is split evenly between two sections.
- Section I: 40 multiple choice questions, 90 minutes, worth 50 percent of your composite score.
- Section II: 6 free response questions, 90 minutes, worth 50 percent. Five of the FRQs are short answer (about 12 to 15 minutes each) and the sixth is the investigative task (about 25 minutes), which counts double.
The composite is curved every year. Recent cutoffs land roughly at 70 plus for a 5, 57 to 69 for a 4, 44 to 56 for a 3. The exam is digital in 2026 through the Bluebook app, so spend at least two hours practicing in Bluebook before test day so the interface is not a surprise.
10 Week Study Timeline
This plan assumes around 60 to 90 minutes of focused study per day, six days per week, layered on top of your class. If you are reviewing only and have already finished the course, compress weeks 1 to 6 into three weeks of unit recall and spend the rest of the time on full timed sections.
Weeks 1 to 2: Units 1 to 3 (Exploring Data and Sampling)
Cover one variable and two variable data, sampling methods, and study design. Most of the foundational vocabulary lives here. Build flashcards for: parameter vs statistic, observational study vs experiment, simple random sample, stratified, cluster, systematic, voluntary response, convenience, undercoverage, response bias, nonresponse bias. Do 30 multiple choice questions per week from a study guide.
Weeks 3 to 4: Units 4 to 5 (Probability and Sampling Distributions)
This is where students lose the most points. Drill the rules: addition rule, multiplication rule, conditional probability, independence, mutually exclusive, expected value, variance of independent random variables. Then push into sampling distributions for proportions and means, including the Central Limit Theorem.
Weeks 5 to 6: Unit 6 (Inference for Proportions)
Confidence intervals and significance tests for one and two proportions. Memorize the four step inference template (covered below) and write out at least 20 full FRQs by hand. The same template repeats for the next three units, so investing time here pays off four times over.
Week 7: Unit 7 (Inference for Means)
One sample t, two sample t, and matched pairs. Remember that matched pairs is just a one sample t test on the differences. Practice identifying which test the prompt requires before you compute anything.
Week 8: Units 8 to 9 (Chi Square and Inference for Slopes)
Chi square goodness of fit, chi square test for homogeneity, chi square test for independence. Then linear regression inference. Many students underprepare for these because they appear at the end of class. They show up reliably on the exam.
Week 9: Mixed Practice and Released FRQs
Take two full released exams under timed conditions. Score them honestly using the College Board scoring guidelines. Build a mistake log: question type, what you missed, the rule that would have saved you. Read the log every morning.
Week 10: Light Review and Test Day
Cut new content. Do one timed FRQ per day plus 15 multiple choice. Sleep on schedule. The night before, scan your mistake log and your inference template card. Stop studying by 8 pm.
All Nine Units at a Glance
Unit 1: Exploring One Variable Data
Center, spread, shape, outliers. Mean vs median, standard deviation vs IQR, dotplots, stemplots, histograms, boxplots. Z scores and percentiles. Know when each summary statistic is appropriate. Skewed data uses median and IQR. Roughly symmetric uses mean and standard deviation.
Unit 2: Exploring Two Variable Data
Scatterplots, direction, form, strength, correlation coefficient r, residuals, least squares regression, coefficient of determination r squared, influential points, leverage, extrapolation. Be ready to interpret slope and intercept in context every single time. Context is graded.
Unit 3: Collecting Data
Sampling vs census, random sampling methods, experiments vs observational studies, control, randomization, replication, blocking, blinding, double blind, placebo. Know how to design a completely randomized design and a randomized block design from a prompt.
Unit 4: Probability, Random Variables, and Probability Distributions
Sample space, complementary events, conditional probability, independence, expected value, variance, binomial distribution, geometric distribution. Memorize binompdf, binomcdf, geometpdf, geometcdf and know which one the question demands.
Unit 5: Sampling Distributions
Sampling distribution of a sample proportion p hat with mean p and standard deviation sqrt(p(1 minus p) over n). Sampling distribution of a sample mean x bar with mean mu and standard deviation sigma over sqrt(n). Central Limit Theorem says the sampling distribution of x bar is approximately normal when n is at least 30, regardless of population shape.
Unit 6: Inference for Categorical Data (Proportions)
One sample z interval and z test for proportions, two sample z interval and z test for proportions. Conditions: random, 10 percent rule, large counts (np and n(1 minus p) at least 10).
Unit 7: Inference for Quantitative Data (Means)
One sample t interval and t test, two sample t interval and t test, matched pairs t. Conditions: random, 10 percent, normal or large sample (n at least 30 by CLT, or sample shows no strong skew or outliers).
Unit 8: Inference for Categorical Data (Chi Square)
Chi square goodness of fit (one categorical variable, one population), chi square homogeneity (one variable, multiple populations), chi square independence (two variables, one population). Conditions: random, expected counts at least 5.
Unit 9: Inference for Quantitative Data (Slopes)
Confidence interval and t test for the slope of the regression line. Conditions: linear, independent, normal residuals, equal variance, random sample. Use LINER as the mnemonic.
Calculator Strategy: TI 84 vs Desmos
You can bring a graphing calculator and you also have access to the Desmos calculator built into Bluebook. The smart approach is to use both for what each does best.
- TI 84 strengths: stat menu (1 Var Stats, 2 Var Stats, LinReg), distribution menu (normalcdf, invNorm, tcdf, binompdf, binomcdf), and the test menu (1 PropZTest, 2 PropZTest, 1 PropZInt, TTest, 2 SampTTest, etc).
- Desmos strengths: graphing functions, fast scatterplots, sliders for what if questions. Use Desmos for problems that would burn time on a TI.
One non negotiable rule: when you use a calculator function on the FRQ, name the function and list the inputs. Writing “1 PropZTest with x equals 47, n equals 200, p sub 0 equals 0.20” earns the methodology point that “I used my calculator and got p equals 0.03” does not.
FRQ Templates That Earn Full Credit
The Four Step Inference Template
Every confidence interval and significance test FRQ uses the same four steps. Memorize the labels. Graders look for them.
For a Significance Test
- State. Define the parameter in context. State the null and alternative hypotheses with the parameter symbol. Set alpha if not given.
- Plan. Name the procedure (one sample t test, two sample z test, etc). Check the conditions explicitly with the data from the prompt, not just by name.
- Do. Compute the test statistic and p value. Show the calculator function used.
- Conclude. “Because p equals 0.03 is less than alpha equals 0.05, we reject H sub 0. We have convincing evidence that…” Always restate the alternative in context.
For a Confidence Interval
- State. Define the parameter and confidence level.
- Plan. Name the procedure. Check conditions.
- Do. Compute the interval.
- Conclude. “We are 95 percent confident that the true (parameter in context) is between A and B.”
Interpretation Sentences You Should Memorize
- Slope: “For each additional one (unit) increase in (x), the predicted (y) increases by (slope) (units).”
- r squared: “(r squared) percent of the variation in (y) is explained by the linear regression of (y) on (x).”
- Confidence interval: “We are C percent confident that the true (parameter) is between A and B.”
- Confidence level: “If we took many samples and constructed a confidence interval each time, about C percent of those intervals would capture the true (parameter).”
- P value: “Assuming H sub 0 is true, there is a (p value) probability of observing a sample result at least as extreme as the one we got.”
- Type I error: “Rejecting H sub 0 when H sub 0 is actually true.”
- Type II error: “Failing to reject H sub 0 when H sub a is actually true.”
Sample Practice Problems
Question 1: Probability
A drug test correctly identifies users 95 percent of the time and correctly identifies non users 90 percent of the time. In a population where 4 percent of people use the drug, what is the probability that a randomly selected person who tests positive is actually a user?
Solution. Use Bayes via a tree. P(user and positive) equals 0.04 times 0.95 equals 0.038. P(non user and positive) equals 0.96 times 0.10 equals 0.096. P(user given positive) equals 0.038 over (0.038 plus 0.096) equals 0.038 over 0.134, which is about 0.284, or 28.4 percent. The classic counterintuitive Bayes result. Even with a 95 percent sensitive test, most positives are false positives in a low prevalence population.
Question 2: One Sample t Test
A coffee shop claims its grande latte contains 16 ounces. A consumer group samples 25 lattes and finds a mean of 15.7 ounces with standard deviation of 0.6 ounces. Is there convincing evidence at the 0.05 level that the lattes contain less than 16 ounces on average?
Solution. State: mu equals true mean ounces. H sub 0: mu equals 16, H sub a: mu less than 16. Plan: one sample t test. Conditions: random sample stated, 10 percent rule (assume more than 250 lattes brewed, reasonable), and n equals 25 with no extreme skew assumed. Do: t equals (15.7 minus 16) over (0.6 over sqrt(25)) equals minus 0.3 over 0.12 equals minus 2.5. df equals 24. p value approximately 0.0098. Conclude: Because p equals 0.0098 is less than alpha equals 0.05, we reject H sub 0. We have convincing evidence the true mean ounces is less than 16.
Question 3: Chi Square Independence
A school surveys students on whether they prefer in person, hybrid, or fully online learning, broken out by grade level (9 through 12). A chi square test of independence yields a test statistic of 18.4 with 6 degrees of freedom. The critical value at alpha equals 0.05 is 12.59. What is the conclusion?
Solution. Because 18.4 is greater than 12.59, we reject H sub 0. There is convincing evidence at the 0.05 level that learning preference and grade level are not independent at this school.
Test Day Strategy
- Pacing. Multiple choice: 90 minutes for 40 questions equals about 2 minutes 15 seconds per question. FRQ: 12 to 15 minutes per short answer, 25 minutes for the investigative task.
- Mark and move. If a multiple choice question takes more than three minutes, flag it and come back. Easy points are scattered throughout the section.
- Always answer in context. Generic statistical language is worth half credit at best. Use the variable names and the units from the prompt in every interpretive sentence.
- Show conditions explicitly. “Random sample stated in problem. 10 percent condition: 25 less than 10 percent of all lattes brewed. Normality: n equals 25 and the problem states no outliers.” Naming alone is not enough.
- Hit the investigative task even if you are unsure. Partial credit is generous. Write the relevant template even if your computation is incomplete.
- Bluebook setup. Practice in Bluebook ahead of time. Know how to access the formula sheet and tables and how to navigate between FRQ parts without losing your work.
Common Mistakes to Avoid
- Forgetting context on every interpretation. The most common single point loss across the entire exam.
- Confusing parameter and statistic. mu, p, beta are parameters. x bar, p hat, b are statistics. Use the right one in your hypotheses.
- Saying “accept H sub 0.” Never. The correct phrasing is “fail to reject H sub 0.”
- Skipping conditions. Even if obviously satisfied, write them out and check with the data.
- Misinterpreting the p value. P value is not the probability that H sub 0 is true. It is the probability of the observed (or more extreme) data given H sub 0 is true.
- Using mean and standard deviation on skewed data. Match the summary to the shape.
- Bare calculator output on FRQs. Always name the function and list inputs.
- Running out of time on the investigative task. Budget 25 minutes. The first parts are usually doable from earlier units.
FAQ
How hard is AP Statistics compared to AP Calculus?
The math is easier. The reasoning, vocabulary, and writing are harder. AP Stats rewards careful readers and clean writers. AP Calc rewards algebra and procedural fluency. Many students do well in one and struggle with the other.
Do I need a TI 84 if Bluebook has Desmos?
Strongly recommended. The TI 84 has dedicated stat tests that would otherwise require setting up by hand. You can use both, and most students who score 5s do.
What is the most common reason students miss a 5?
Inadequate FRQ writing. Two students with the same answers can earn very different scores based on whether they used the four step template, named procedures, checked conditions, and answered in context.
How important is the formula sheet?
It contains the formulas you need. Spend an afternoon practicing locating each one quickly. Knowing where the chi square formula lives saves 30 seconds on the exam, and 30 seconds adds up.
How do I prepare for the investigative task?
Review every released investigative task for the last six years. They reuse problem types: experimental design, simulation, two way tables, and unfamiliar inference settings. Patterns recur.
Can I self study AP Stats?
Yes. The College Board CED is freely available. Pair it with a video resource (Mr Roush, jbstatistics on YouTube) and a question bank. Self studiers regularly earn 5s.
Take a Free AP Statistics Practice Test
The fastest way to spot weak units is to sit a timed multiple choice block and review every miss with the four step template. Take our free AP Statistics practice tests at PracticeTestVault and build your study plan from the units that bleed points. Pair this guide with our AP Calculus AB study guide, our AP Biology study guide, and our guide on reviewing practice question rationales to make every minute of prep count.
The 5 on AP Statistics is not a function of being a math genius. It is a function of practicing the templates, writing in context, and checking conditions on every single inference question. Ten focused weeks gets you there.