How to study for AP Statistics: a real plan

Study for the exam you are actually taking

The AP Statistics exam is 3 hours long and splits into two equally weighted halves. Section I is 42 multiple-choice questions in 90 minutes. Section II is four free-response questions in 90 minutes, each worth 10 points and 12.5% of your score.

The four free-response questions arrive in a fixed order, and knowing it changes how you prepare:

1. Question 1 focuses on Practices 1 and 2: formulating a statistical question and collecting data.
2. Question 2 focuses on Practices 3 and 4: analyzing data and interpreting results.
3. Question 3 is inference, either a hypothesis test or a confidence interval.
4. Question 4 spans Practices 2, 3, and 4 across several content areas.

A graphing calculator with statistical capabilities is expected in both sections, and formulas and tables are provided in both. Notice what that order implies: only Question 3 is designated the inference question, so a plan built on computation alone aims at a quarter of the section. Full detail sits on the exam format page.

Weight your review by the exam weightings

The multiple-choice section has published unit weightings, and the Course and Exam Description suggests pacing in 45-minute class periods.

Two readings are worth acting on. Unit 1 carries the largest multiple-choice weight, so a shaky Unit 1 leaks points across the whole section. Unit 5 gets about 9 class periods for the same weight Unit 4 gets 18 for, which makes regression the unit most likely to have been rushed in class and the cheapest to repair alone.

Unit pages to work from: Unit 1, Unit 2, Unit 3, Unit 4, and Unit 5.

Memorize sentences, not formulas

The formula sheet is provided in both sections; the sentences are not. That fact should reorganize your revision. Every hour spent copying formulas onto flashcards is an hour not spent on the four templates below.

Confidence interval. "We are $C\%$ confident that the interval from [low] to [high] captures the true [parameter, in context]." The confidence level describes the method across many samples, not the one interval in front of you. See what 95% confidence means.

Test conclusion. "Because the p-value of [value] is [less than or greater than] $\alpha =$ [level], we [reject or fail to reject] $H_0$. There [is or is not] convincing evidence that [alternative hypothesis, in context]." You never accept $H_0$. See what a p-value means.

Slope. "For each additional one [x-unit], the predicted [y-variable] changes by [slope] [y-units]." The word predicted is doing real work there.

Coefficient of determination. "About [$r^2$]% of the variation in [y-variable] is explained by the linear model with [x-variable]." Here $r^2$ is r-squared, the square of the correlation. See r vs r-squared.

Memorize one more rule alongside them: hypotheses are written about a parameter such as $p$ (the population proportion) or $\mu$ (mu, the population mean), never about a sample statistic like $\hat{p}$ (p-hat, the sample proportion) or $\bar{x}$ (x-bar, the sample mean). See null vs alternative hypothesis.

Drill conditions until you write them with numbers

The Course and Exam Description says students routinely fail to earn condition points because they check conditions superficially, without connecting them to the problem in front of them. Writing "SRS" is not a check. Writing "a simple random sample of 60 students was taken, and 60 is less than 10% of the 1,400 students at the school" is.

Keep this list somewhere you see it daily:

• Random. Name the source of randomness: a random sample from a population, or random assignment to treatments.
• 10%. When you sample without replacement, the sample size $n$ is at most 10% of the population. Write the comparison out.
• Large counts (proportions). $np$ and $n(1-p)$ are each at least 10. Use the null value $p_0$ for a test and $\hat{p}$ for an interval.
• Normal or large sample (means). The population is stated to be normal, or $n \ge 30$, or a graph of the sample shows no strong skew and no outliers.
• Expected counts (chi-square). Every expected count is at least 5, and it is the expected counts you check, not the observed ones.

Drill these against real problems rather than reciting them. The proportion z-test practice set and the t-test practice set both make you state conditions before computing.

An eight-week sequence

This assumes about 5 hours a week and that you have seen the material once in class. Halve it to four weeks by doubling up.

1. Week 1, one-variable data. Shape, center, spread, unusual features, and the median-excluded (TI-84) quartile rule this site uses throughout. Work describing distributions practice.
2. Week 2, collecting data. Sampling methods, sources of bias, and experimental design, which is Question 1 territory on the free response. Work sampling methods practice and experimental design practice.
3. Week 3, probability. The addition and multiplication rules, conditional probability, and independence. Work probability rules practice and conditional probability practice.
4. Week 4, random variables and the normal model. Expected value, the binomial distribution, z-scores, and then sampling distributions, which is the pivot into all of inference. Work normal distribution practice and sampling distribution practice.
5. Week 5, inference for proportions. One-proportion and two-proportion procedures plus chi-square for two-way tables, writing every conclusion out longhand. Work confidence interval practice and two-proportion inference practice.
6. Week 6, inference for means. One-sample, paired, and two-sample $t$ procedures, and knowing which one a scenario calls for. Work t-test practice and two-sample t practice.
7. Week 7, regression and mixed review. Scatterplots, correlation, the least-squares line, residual plots, then a mixed set that forces you to choose a procedure. Work regression practice and read which statistical test to use.
8. Week 8, full timed run and repair. One complete timed exam, then spend the rest of the week only on what you got wrong.

Inside each week the pattern is the same: compute, write the interpretation from memory, check it against the template.

The final week checklist

In the last seven days, stop learning new content and start protecting points you already have.

1. Write the four interpretation templates from memory, on paper, twice.
2. Write the condition list from memory and check it against the list above.
3. Do one timed 42-question multiple-choice section against a 90-minute clock.
4. Do four free-response questions under a 90-minute clock, in exam order, without looking anything up.
5. Score your own free response harshly: no context means no point.
6. Re-read the cram sheet once, slowly, the day before.
7. Confirm you can find any value fast on the formula sheet, the z-table, the t-table, and the chi-square table.
8. Practice the calculator sequences you use, remembering that a bare calculator command is not an answer.
9. Skim the FRQ guide for the point-by-point scoring pattern.
10. Sleep. Section II is 90 minutes of careful writing.

How to practice writing when nobody is grading you

Most students never get enough graded free response, so grade yourself. After each inference answer, ask four questions and take a point off for every miss.

• Did you name the exact procedure, in words or by formula?
• Did you verify each condition with numbers from the problem?
• Did you show the statistic, the test statistic or interval, and the p-value?
• Did you conclude in context, linked to the p-value or the interval?

Then apply the stranger test to your last sentence. If someone who never read the problem could not tell what was being studied, that is a conclusion without context, and a lost point.

Rewriting beats re-reading. Take an answer you scored 2 out of 4 on and rewrite it to full credit the same day. The FRQ guide has worked examples, and the flashcards cover the vocabulary those sentences lean on.

What not to spend time on

Some study effort is wasted, and cutting it frees hours for writing practice.

Memorizing what is printed. Formulas and tables are provided in both sections. Learn to read the sheet quickly instead, using the formula sheet.

Topics the redesign removed. Chi-square goodness-of-fit, the geometric distribution, inference for a regression slope, combining random variables, and transformations to achieve linearity are not assessed on this exam. They remain standard college content, so learn them if you are taking such a course, but keep them off an AP revision plan. Binomial vs geometric explains the distinction and flags the removal.

Bare calculator output. The Course and Exam Description expects a calculation to show an expression, the values substituted into it, and an answer, so calculator results with no supporting work do not earn full computation credit. See reading calculator output.

When you need a number checked fast, the calculators do the arithmetic so you can spend the time on the sentence. The official course page is at AP Central.