Chi-Square Tests in AP Statistics

In AP Statistics, chi-square tests are inference procedures for categorical data that measure how far observed counts in a two-way table fall from expected counts. They test homogeneity (do two or more populations share one distribution?) or independence (are two variables associated in one population?).

What are Chi-Square Tests?

A chi-square test checks whether the counts you actually observed in a table are close to the counts you'd expect if the null hypothesis were true. The chi-square statistic measures the distance between observed and expected counts relative to the expected counts (3.14.A.1). Think of it as a "surprise meter." A small value means the data look like what H₀ predicts. A big value means the data are hard to explain if H₀ is true.

The statistic follows a chi-square distribution. These distributions only take positive values and are skewed right. The skew gets less extreme as the degrees of freedom increase (3.14.A.2). Topic 3.14 covers two versions for two-way tables. A test for homogeneity asks whether the distribution of one categorical variable is the same across two or more populations or treatments (3.14.B.1). A test for independence asks whether two categorical variables are associated within a single population (3.14.B.3). Same math, different question, different data-collection setup.

Why Chi-Square Tests matter in AP® Statistics

Chi-square tests live in Topic 3.14 (Setting Up a Chi-Square Test for Homogeneity or Independence) in Unit 3: Inference for Categorical Data. They support four learning objectives. AP Stats 3.14.A covers describing chi-square distributions. AP Stats 3.14.B covers choosing the right test and naming the populations and variables. AP Stats 3.14.C covers writing the hypotheses. AP Stats 3.14.D covers justifying the test by checking conditions. This is where proportion inference goes past two groups. A two-proportion z-test can compare two groups on one yes/no outcome. A chi-square test can compare several groups across several categories at once. The exam rewards you for picking the right procedure, so you need to know when a z-test isn't enough and a chi-square test is.

How Chi-Square Tests connect across the course

Homogeneity vs. Independence (Unit 3)

These are the two chi-square tests for two-way tables, and the calculation is identical for both. The difference is the sampling design. Separate samples from several populations mean homogeneity. One sample classified on two variables means independence. Figure out how the data were collected and you know which test you're running.

Expected Frequencies and Two-Way Tables (Units 1, 3)

You first built two-way tables and compared conditional distributions in Unit 1. The chi-square test turns that eyeballing into formal inference. Each expected count is the count you'd see if the conditional distributions matched perfectly: (row total × column total) ÷ table total.

Independent Events in Probability (Unit 2)

In Unit 2, events are independent when knowing one doesn't change the probability of the other. The chi-square test for independence applies that same idea to whole variables. H₀ says the row variable tells you nothing about the column variable, and the test checks whether the data are consistent with that.

Degrees of Freedom (Units 3, 4)

Degrees of freedom pick which chi-square curve you use, just as they pick which t-distribution you use in Unit 4. For a two-way table, df = (rows − 1)(columns − 1). More df means a less skewed curve, so the same statistic can be unusual at low df and ordinary at high df.

Are Chi-Square Tests on the AP® Statistics exam?

Chi-square tests appear on the exam mostly as setup and justification questions. You'll be asked to:

• Pick the right test. Decide between homogeneity and independence from the study design, then name the populations and the categorical variable(s) in context (3.14.B).
• Write hypotheses in words. For homogeneity, H₀ says there's no difference in distributions across the populations or treatments. For independence, H₀ says there's no association between the variables (3.14.C).
• Check conditions. You need randomization, the 10% condition (skip it for a randomized experiment), and all expected counts greater than 5 (3.14.D.1).

Multiple-choice items like to test the edge cases. One common stem gives a 4×4 table (exercise frequency vs. sleep quality) from only 200 adults and asks which situation would violate conditions. With 16 cells, small expected counts are a real risk. Another stem describes a stratified random sample and asks whether randomization is still met. A third describes students randomly assigned to teaching methods, which makes it an experiment, so the 10% condition doesn't apply. No released FRQ has used the term verbatim. Still, the full inference structure FRQs reward (hypotheses, conditions, conclusion in context) applies directly to chi-square tests.

Chi-Square Tests vs Chi-square test for homogeneity vs. chi-square test for independence

Both use the same statistic, expected counts, and df, so the mix-up comes from the design, not the math. Homogeneity uses separate random samples (or treatment groups) from two or more populations and asks whether one categorical variable has the same distribution in each. Independence uses one random sample from a single population where each individual is classified on two categorical variables, and it asks whether those variables are associated. A quick check is to ask whether the researcher chose how many people go in each group. If yes, it's usually homogeneity.

Key things to remember about Chi-Square Tests

• A chi-square test measures how far observed counts are from expected counts, relative to the expected counts, so bigger values give stronger evidence against H₀.

• Use a chi-square test for homogeneity when comparing the distribution of one categorical variable across two or more populations or treatments.

• Use a chi-square test for independence when one sample is classified on two categorical variables and you want to know if they're associated.

• The conditions are random sampling or random assignment, the 10% condition unless it's a randomized experiment, and all expected counts greater than 5.

• Chi-square distributions are always positive and skewed right, and they become less skewed as the degrees of freedom increase.

• Always state hypotheses and conclusions in context, naming the specific populations and categorical variables in the study.

Frequently asked questions about Chi-Square Tests

What is a chi-square test in AP Statistics?

It's an inference procedure for categorical data that compares observed counts in a two-way table to the counts expected under H₀. In Topic 3.14 you use it to test homogeneity across populations or independence between two variables.

How is a chi-square test for homogeneity different from a test for independence?

Homogeneity compares one variable's distribution across two or more separately sampled populations or treatment groups. Independence uses one sample from one population and checks whether two variables are associated. The calculations are identical, so the study design decides which one you're doing.

Do all the counts in a chi-square test have to be greater than 5?

No. The condition applies to the expected counts, not the observed counts. Every expected count must be greater than 5, and an observed count of 2 is fine as long as its expected count clears 5.

Do you need the 10% condition for a chi-square test?

Only when you sample without replacement from a population. If subjects were randomly assigned to treatments, as in a study where students are randomly assigned to Traditional, Flipped, or Online teaching, it's an experiment and the 10% condition doesn't apply.

How do you find degrees of freedom for a chi-square test?

For a two-way table, df = (number of rows − 1)(number of columns − 1). A 3×4 table gives df = 2 × 3 = 6. Higher df gives a chi-square curve that is less skewed right.