What was removed from AP Statistics

By Jude Wallis · Published August 11, 2026

College Board removed five entries: analyzing departures from linearity, combining random variables, the geometric distribution, chi-square goodness of fit, and the old Unit 9 on slopes. What went is transforming data to straighten a curve; reading curvature off a residual plot is still assessed.

The topics removed from AP Statistics by the Fall 2026 redesign, quoted from the College Board future revisions page, with the surviving content located by topic code in the current Course and Exam Description.

The removal list, as College Board published it

One sentence on the College Board AP Statistics future revisions page carries the whole change: the following topics were removed: analyzing departures from linearity (2.9), combining random variables (4.9), the geometric distribution (4.12), chi-square goodness of fit test (8.2 and 8.3), and Unit 9: Inference for Quantitative Data: Slopes.

Five entries: five old topic codes and one whole unit. That is the complete published list, and this page adds nothing to it.

It is worth being clear about where this does and does not come from. The Course and Exam Description effective Fall 2026 is a course framework: it describes the course that exists, not the one that used to. It says nothing about removals, so the revisions page is the source for this list, and it is the page to cite if a teacher asks where you got it.

The rest of this guide does three things: translates the old topic numbers, so you do not cut the wrong material; explains what each removed entry actually covered; and handles the one removal that is routinely over-read.

Those topic numbers are OLD numbers

The codes in the removal list index the nine-unit course. The Fall 2026 course reuses the same shape of code for different content, and two of the six collide with topics that are very much still examined.

Code in the removal list	What it was, in the old course	What that code means in the Fall 2026 course
2.9	Analyzing departures from linearity	Parameters of Random Variables, in Unit 2
4.9	Combining random variables	Setting Up a Test for the Difference Between Two Population Means, in Unit 4
4.12	The geometric distribution	Nothing. Unit 4 ends at topic 4.10
8.2 and 8.3	Chi-square goodness of fit	Nothing. The Fall 2026 course has five units, so there is no Unit 8
Unit 9	Inference for Quantitative Data: Slopes	Nothing. There is no Unit 9

So if you take the removal list at face value against a current syllabus, you would delete two live topics and hunt for three that do not exist. Check titles, never numbers. The current codes and titles are on the unit pages and on the units guide.

This is also the single most common way a well-meaning resource gets the redesign wrong: it quotes the removal list correctly and then indexes it against the new framework.

Analyzing departures from linearity, and the part that stayed

This is the removal students and teachers most often over-read, so it deserves the careful sentence.

The old topic 2.9 was about re-expressing data to straighten a curve: taking logs of $y$ , taking logs of both variables, taking a square root, then fitting a line to the transformed data and back-transforming a prediction. That is what left. The Fall 2026 course has no topic on transforming data to achieve linearity, and you will not be asked to produce or interpret a log-log model.

What did not leave is noticing that a line is wrong in the first place. Topic 5.4 in Unit 5 is Residuals, and a curved pattern in a residual plot is exactly the evidence that a linear model does not fit. You are still expected to look at a residual plot and say so.

Write it this way and you will always be right: the course has no topic on transforming data to achieve linearity. Do not write that departures from linearity are not tested, because detecting them is a live skill in Unit 5.

The worked example below fits a line to curved data and reads the residuals. For more, see how to interpret residual plots, the glossary entry on transformation to achieve linearity, and residual plots practice.

Chi-square goodness of fit went, homogeneity and independence stayed

The old topics 8.2 and 8.3 covered the chi-square goodness of fit test: one categorical variable, a set of observed counts in a single row, and a claimed distribution to test them against. That test is off the exam.

The other two chi-square tests are not. In the Fall 2026 course they sit in Unit 3 as topics 3.14, Setting Up a Chi-Square Test for Homogeneity or Independence, and 3.15, Carrying Out a Chi-Square Test for Homogeneity or Independence. Both work from a two-way table.

The practical rule for telling which is which has not changed, and it is now the rule for telling what is examinable at all:

One-way table, counts of one categorical variable, compared to a claimed distribution. Goodness of fit. Not on this exam.
Two-way table, one sample classified by two variables. Test for independence. Topics 3.14 and 3.15.
Two-way table, several populations or treatment groups compared on one variable. Test for homogeneity. Topics 3.14 and 3.15.

The expected count for a two-way table cell is the row total times the column total divided by the grand total, and the worked example below runs one. The condition to check is that every expected count is at least 5, and it is the expected counts you check, not the observed. See chi-square tests explained, the chi-square test glossary entry, chi-square two-way practice and Unit 3 in full.

The geometric distribution went, the binomial stayed

The old topic 4.12 was the geometric distribution: keep repeating an independent trial until the first success, and let the random variable be the number of trials that takes. It is off the exam.

The binomial distribution is topic 2.10 in the Fall 2026 course and is fully examinable: its conditions, its probabilities, its mean and its standard deviation.

The distinction is worth keeping straight anyway, because it is how you recognize a binomial setting in the first place. In a binomial setting the number of trials is fixed in advance and the number of successes varies. In a geometric setting the number of successes is fixed at one and the number of trials varies. If a question says keep going until, that is the shape of a geometric problem, and on this exam it will not be asked as one.

The geometric distribution remains standard content in a college introductory statistics course, so learn it if that is where you are headed. For an AP revision plan, spend the time on the binomial distribution instead, and see binomial vs geometric for the comparison, how to tell if a situation is binomial for the recognition drill, and Unit 2 in full for where it now sits.

Combining random variables

The old topic 4.9 covered combining random variables: sums and differences of independent random variables, and the rules for the mean and the standard deviation of the result. That entry is on the removal list.

Random variables themselves are not. Unit 2 of the Fall 2026 course runs Introduction to Random Variables and Probability Distributions at topic 2.8 and Parameters of Random Variables at topic 2.9, so expected value and standard deviation of a single random variable are live content. See how to find expected value and is expected value the same as the mean?.

One point of confusion to head off. Removing this topic does not remove inference for a difference, which is a different idea in a different place. Unit 3 tests the difference between two proportions at topics 3.9 to 3.13, and Unit 4 tests the difference between two means at topics 4.6 to 4.10. Those procedures work from sampling distributions of statistics, and they occupy ten of the twenty-five topics in the two inference units. If a review chapter is titled combining random variables, you can skip it; if it is titled two-sample inference, you cannot.

A whole unit went: inference for slopes

The largest single removal is the old Unit 9, Inference for Quantitative Data: Slopes. There is no confidence interval for a regression slope and no significance test for a regression slope on this exam, and the $t$ procedures built on the standard error of the slope are gone with them.

Regression is still a unit. Unit 5, Regression Analysis keeps five topics, none of them inferential, and the units guide lists them with their codes.

So what you still need from regression is the descriptive and diagnostic side, which is most of what students find hard about it anyway: describing a scatterplot, interpreting $r$ and $r^2$ , writing the equation of the least-squares line, interpreting the slope and the intercept in context, making a prediction and knowing when not to, and reading a residual plot.

This one has an obvious practical consequence for older material. Computer regression output tables, the ones with a column of standard errors and $t$ statistics and p-values for the slope, still appear in old practice sets as inference questions. On this exam that output is a source of the slope and the intercept, not the input to a hypothesis test about the slope. Work regression practice and regression output practice rather than the old slope-inference sets, and see how to interpret the slope of a regression line.

What to do with the removed material

Removed from the AP exam does not mean wrong or useless. Most of these five are standard content in a college introductory statistics course, and three of them (goodness of fit, the geometric distribution, inference for slopes) are things you will meet again immediately if you take one.

For the AP exam, the rule is simple. Do not revise them, do not lose confidence when a practice question uses them, and do not assume a source is worthless because it teaches them.

A cheap audit of anything you already own: scan the contents page for a chapter on inference for the slope of a regression line, a chapter or section on goodness of fit, and a section on the geometric distribution. Their presence tells you the book was written for the nine-unit course, which tells you the structural material in it is out of date even where the statistics is fine. Should I use old AP Statistics review books? works through that audit, and what changed in AP Statistics for 2027 covers the redesign as a whole.

The removal list is only half of the change. The other half is the exam itself, which becomes fully digital in May 2027: see is the AP Statistics exam digital?. For what the course does contain, the course hub and the units guide have every unit and topic.

Curvature in a residual plot, which is still assessed

A student fits a least-squares line to these five points: $(1, 2)$ , $(2, 5)$ , $(3, 10)$ , $(4, 17)$ , $(5, 26)$ . Find the line, compute the residuals, and say what the residual plot shows. Is this skill on the Fall 2026 exam?

Find the means: $\bar{x} = \frac{1+2+3+4+5}{5} = 3$ and $\bar{y} = \frac{2+5+10+17+26}{5} = \frac{60}{5} = 12$ .
Compute $S_{xx} = \sum (x - \bar{x})^2 = 4 + 1 + 0 + 1 + 4 = 10$ .
Compute $S_{xy} = \sum (x - \bar{x})(y - \bar{y}) = (-2)(-10) + (-1)(-7) + (0)(-2) + (1)(5) + (2)(14) = 20 + 7 + 0 + 5 + 28 = 60$ .
Slope: $b = \frac{S_{xy}}{S_{xx}} = \frac{60}{10} = 6$ . Intercept: $a = \bar{y} - b\bar{x} = 12 - 6(3) = -6$ . So $\hat{y} = -6 + 6x$ .
Predicted values at $x = 1$ through $5$ : $0$ , $6$ , $12$ , $18$ , $24$ .
Residuals, observed minus predicted: $2 - 0 = 2$ , $5 - 6 = -1$ , $10 - 12 = -2$ , $17 - 18 = -1$ , $26 - 24 = 2$ .
Check they sum to zero, as least-squares residuals must: $2 - 1 - 2 - 1 + 2 = 0$ .
Plot the residuals against $x$ and read the pattern: positive, negative, negative, negative, positive. That is a U shape, not random scatter, so a line is the wrong model for these data.

The line is $\hat{y} = -6 + 6x$ and the residuals are $2, -1, -2, -1, 2$ , a clear U-shaped pattern. Yes, this is examinable: reading curvature in a residual plot is topic 5.4. What was removed is the next step the old course took, re-expressing the data to straighten the curve.

An expected count for the chi-square test that stayed

A survey of 120 students cross-classifies year group against a yes or no answer. One cell sits in a row with a total of 60 and a column with a total of 50. Find that cell's expected count, and say which chi-square test this is.

Identify the table. Two categorical variables recorded on one sample of 120 students gives a two-way table, so this is a chi-square test for independence, topics 3.14 and 3.15.
Use the expected count formula: expected $= \frac{\text{row total} \times \text{column total}}{\text{grand total}}$ .
Substitute: $\frac{60 \times 50}{120} = \frac{3000}{120} = 25$ .
Interpret it: if year group and answer were independent, you would expect about 25 of the 120 students in that cell.
Check the condition on this cell: the expected count of 25 is at least 5, so this cell is fine. Every expected count in the table has to clear 5, and it is the expected counts you check, not the observed ones.
Note the contrast with the removed test: goodness of fit works from a single row of counts and a claimed distribution, and its expected counts are $np_i$ rather than a row-by-column product. That test is not on this exam.

The expected count is 25. This is a chi-square test for independence, which is still examinable at topics 3.14 and 3.15. The goodness of fit test, old topics 8.2 and 8.3, is the one that was removed.

Frequently asked questions

Is the chi-square goodness of fit test on the AP Statistics exam?

No. College Board's revisions page lists it among the removals, as old topics 8.2 and 8.3. The chi-square tests for homogeneity and independence are still examinable, as topics 3.14 and 3.15 in Unit 3. Both of those work from a two-way table, which is the quickest way to tell them apart from goodness of fit.

Are departures from linearity still tested?

Detecting them is. What was removed is the old topic on re-expressing data to straighten a curve, so the course has no topic on transforming data to achieve linearity. Curvature in a residual plot is still the evidence that a linear model is wrong, under topic 5.4, Residuals.

Is the geometric distribution still on the AP Statistics exam?

No. It was removed as old topic 4.12. The binomial distribution stayed and is topic 2.10 in Unit 2. The geometric distribution remains standard in a college introductory course, so it is worth knowing if you are going on, but it does not belong in an AP revision plan.

Is inference for the slope of a regression line still tested?

No. The entire old Unit 9, Inference for Quantitative Data: Slopes, was removed, so there is no confidence interval or significance test for a slope on this exam. Unit 5 keeps regression as a descriptive and diagnostic topic: correlation, the least-squares line, prediction, and residual plots.

Why do the removed topic numbers not match my syllabus?

Because they are numbers from the nine-unit course. In the Fall 2026 course, 2.9 is Parameters of Random Variables and 4.9 is Setting Up a Test for the Difference Between Two Population Means, both examinable, and there is no 4.12, no Unit 8 and no Unit 9. Match on titles rather than on codes.