Test Statistic in AP Statistics

In AP Statistics, a test statistic is a standardized number calculated from sample data that measures how far the sample result is from the value claimed in the null hypothesis, usually in standard errors. It is then used to find the p-value.

What is Test Statistic?

A test statistic answers one question. How surprising is my sample if the null hypothesis is true? It takes what you observed, compares it to what H₀ predicts, and puts that gap on a standard scale so you can look it up in a known distribution.

For z and t tests, the recipe is always the same:

test statistic = (sample statistic - null value) / standard error of the statistic

Think of it as a ruler measured in standard errors. A z of 2.5 means your sample landed 2.5 standard errors away from the null value. That's far, so it's evidence against H₀. A value near 0 means your data look about like what H₀ predicts. In AP Stats you'll see three main types: z for proportions (one-sample and two-sample), t for means (one-sample, paired, and two-sample), and χ² (chi-square) for categorical data in goodness-of-fit, homogeneity, and independence tests. The chi-square statistic is built differently, as a sum of (observed - expected)²/expected. It still does the same job of measuring how far the data are from what H₀ expects.

Why Test Statistic matters in AP® Statistics

The test statistic is the middle step of every significance test in the course. It shows up in Topic 3.7 (one-proportion z test), Topic 3.13 (two-proportion z test), Topic 3.15 (chi-square tests for homogeneity or independence), Topic 4.5 (one-sample and paired t tests), and Topic 4.10 (two-sample t test). It's also the bridge to Topic 3.6, because you can't get a p-value without one. The p-value is just the probability of getting a test statistic at least as extreme as yours, assuming H₀ is true. On free-response inference questions, the "do" step usually earns credit only if you name the test, show the test statistic with correct values plugged in, and report the p-value. So the test statistic is where your work moves from setup to evidence.

How Test Statistic connects across the course

P-Value (Unit 3)

The test statistic tells you where your data landed, and the p-value tells you how rare it is to land there or farther out. A bigger test statistic in the direction of Hₐ always means a smaller p-value. That's why a strong z or t leads to rejecting H₀.

Pooled Proportion in Two-Proportion Tests (Unit 3)

In a two-proportion z test (Topic 3.13), H₀ says p₁ = p₂. So you combine both samples into a pooled proportion to compute the standard error in the denominator. The confidence interval in Topic 3.10 doesn't assume the proportions are equal, so it uses the unpooled standard error and a critical value instead of a test statistic.

Chi-Square Statistic (Unit 3)

The χ² statistic is the test statistic for categorical data with more than two categories or groups (Topic 3.15). Instead of one difference divided by a standard error, it adds up the squared gaps between observed and expected counts across every cell. It can never be negative, so bigger simply means more evidence against H₀.

t Statistic for Means (Unit 4)

When you test a mean (Topics 4.5 and 4.10), you don't know σ, so you plug in the sample standard deviation and get a t statistic instead of a z. The formula has the same structure. You just read it from a t-distribution with the right degrees of freedom, which has fatter tails than the normal curve.

Is Test Statistic on the AP® Statistics exam?

Multiple choice. You'll be asked to turn a given test statistic into a p-value, or to connect it to a critical value. For example, with H₀: μ = 85, Hₐ: μ < 85, and z = -1.28, you need the left-tail area, which is about 0.10. Another common stem gives t = -1.96 with 24 degrees of freedom and α = 0.05 for a left-tailed test. The critical value is about -1.711, and -1.96 lies beyond it, so the p-value is less than 0.05 and you reject H₀. Both methods have to agree. Many questions also test whether you can tell the test statistic apart from the p-value it produces.

Free response. Released FRQs like 2017 Q5, 2018 Q4, 2018 Q6, and 2019 Q4 ask whether the data give convincing statistical evidence for a claim. Full credit usually means you (1) identify the correct test, (2) check conditions such as random sampling and the 10% condition, (3) calculate the test statistic with the formula or with labeled calculator values, (4) give the p-value, and (5) state a conclusion in context. If you just write "z = 2.31" with no test named or no values shown, you risk losing the mechanics credit.

Test Statistic vs P-Value

The test statistic is a distance, measured in standard errors (or as a χ² sum). It can be negative, zero, or large. The p-value is a probability between 0 and 1 that comes from the test statistic. As a rough guide, a large |z| or |t| means your data sit far from H₀, and the p-value tells you how unlikely that distance is if H₀ is true. You compare the p-value to α, not the test statistic. The only time you compare the test statistic directly is when you check it against a critical value.

Key things to remember about Test Statistic

• A test statistic measures how many standard errors the sample result is from the null hypothesis value.

• For z and t tests, the formula is (sample statistic minus null value) divided by the standard error.

• Use z for proportions, t for means, and chi-square for counts in categorical data.

• The farther the test statistic is from zero in the direction of the alternative hypothesis, the smaller the p-value and the stronger the evidence against H₀.

• The two-proportion z test uses a pooled proportion in its standard error because the null hypothesis assumes the two population proportions are equal.

• On FRQs, show the test statistic with values substituted or calculator inputs labeled, then report the p-value and a conclusion in context.

Frequently asked questions about Test Statistic

What is a test statistic in AP Statistics?

It's a standardized value calculated from your sample that shows how far the result is from what the null hypothesis claims. For z and t tests it equals (statistic - null value) / standard error, and you use it to find the p-value.

Is the test statistic the same as the p-value?

No. The test statistic is a distance, like z = -1.28, and the p-value is the probability of getting a result that extreme or more if H₀ is true. For z = -1.28 in a left-tailed test, that p-value is about 0.10.

Can a test statistic be negative?

Yes, for z and t tests. A negative value means the sample statistic fell below the null value, which supports a left-tailed alternative like Hₐ: μ < 85. A chi-square statistic can never be negative because it adds up squared differences.

How is a test statistic different from a critical value?

The test statistic comes from your data, and the critical value comes from α and the distribution. With t = -1.96, df = 24, and α = 0.05 (left-tailed), the critical value is about -1.711. Your statistic is beyond it, so you reject H₀, and that matches a p-value below 0.05.

Do I need to show the test statistic formula on the AP Stats FRQ?

You need to show enough work for a reader to follow. That means either the formula with values plugged in, or the test name with labeled calculator inputs. Inference FRQs such as 2018 Q6 reward naming the test, computing the statistic and p-value, and concluding in context.