A test statistic is a standardized value, calculated from sample data, that measures how far your observed result falls from what the null hypothesis predicts. In AP Stats it takes the form (statistic − parameter)/standard error for z- and t-tests, or χ² = Σ(Observed − Expected)²/Expected for chi-square tests.
A test statistic answers one question: "If the null hypothesis were true, how surprising is my sample?" It converts your raw data into a standardized number so you can compare it to a known distribution (z, t, or chi-square) and get a p-value.
For z- and t-tests, the test statistic follows the same skeleton every time: (sample statistic − null parameter value) ÷ standard error of the statistic. That's really just counting how many standard errors your sample result sits from the null value. A one-sample t-test uses t = (x̄ − μ)/(s/√n); a two-sample t-test uses t = ((x̄₁ − x̄₂) − (μ₁ − μ₂))/√(s₁²/n₁ + s₂²/n₂). Chi-square tests use a different recipe, χ² = Σ(Observed − Expected)²/Expected, which measures the total distance between observed counts and the counts the null hypothesis expects, relative to those expected counts. The bigger the test statistic (in absolute value for z and t, or just bigger for χ²), the more your data disagrees with the null, and the smaller your p-value gets.
The test statistic is the engine of every significance test in Units 6, 7, and 8, which together make up the biggest chunk of the AP Stats exam. You calculate one in a one-proportion z-test (Topic 6.6), a one-sample t-test for a mean (LO 7.5.A), a two-sample t-test (LO 7.9.A), a chi-square goodness-of-fit test (LO 8.3.A), and chi-square tests for homogeneity or independence (LO 8.6.A). The CED includes a clarifying statement worth knowing: test statistic formulas are NOT on the AP formula sheet as labeled formulas, but you can rebuild any of them from the general pattern (statistic − parameter)/standard error plus the standard error formulas that ARE on the sheet. The test statistic is also what defines the p-value. Per the CED, the p-value is the probability of getting a test statistic as extreme or more extreme than yours, assuming the null is true. No test statistic, no p-value, no conclusion.
Keep studying AP Statistics Unit 8
P-Value (Units 6-8)
The p-value is literally defined in terms of the test statistic. It's the probability, assuming the null hypothesis is true, of getting a test statistic as extreme or more extreme than the one you observed. The test statistic is the raw measurement; the p-value translates it into a probability you can compare to α.
Null Hypothesis (Units 6-8)
Every test statistic is built around the null. The parameter value you subtract in z = (p̂ − p₀)/SE comes straight from H₀, and the expected counts in a chi-square statistic are the counts H₀ predicts. The test statistic only means something relative to the null distribution.
Chi-Square Statistic (Unit 8)
The chi-square statistic is the test statistic for categorical data with multiple categories. It breaks the (statistic − parameter)/SE mold by summing squared distances between observed and expected counts instead, and its distribution is always right-skewed with only positive values (per LO 8.2.A).
Critical Value (Units 6-8)
A critical value is the cutoff on the same distribution your test statistic lives on. Instead of converting the test statistic to a p-value, you can compare it directly to the critical value for your α. Critical values also show up in confidence intervals, where the same z* or t* sets the margin of error.
Calculating and interpreting a test statistic is a guaranteed skill on the exam. Multiple choice questions hand you a test statistic and ask for the conclusion, like a goodness-of-fit problem where χ² = 9.72 with 4 categories and you have to find df = 3, get the p-value, and compare it to α = 0.05. Free response questions test it inside the full four-step inference procedure. The 2018 FRQs asked for complete significance tests on blood pressure means and ACL surgery recovery, and the 2017 and 2019 FRQs required chi-square reasoning on two-way tables. To earn full credit you have to name the test, check conditions, show the test statistic calculation (or correctly report it from the formula), find the p-value, and write a conclusion in context. A common point-loser is reporting a test statistic with no degrees of freedom for t and chi-square tests, or naming the wrong test so the whole statistic is wrong from the start.
The test statistic measures distance; the p-value measures probability. The test statistic tells you how many standard errors (or chi-square units) your sample sits from the null value. The p-value then asks how likely a value that extreme would be if the null were true. You always compute the test statistic first, then use its distribution (z, t, or chi-square) to find the p-value. Comparing the test statistic itself to α is a classic error, since α is a probability and only the p-value can be compared to it.
A test statistic standardizes your sample result so you can measure how far it falls from what the null hypothesis predicts.
For z- and t-tests, every test statistic follows the same pattern: (sample statistic − null parameter value) divided by the standard error.
Chi-square tests use χ² = Σ(Observed − Expected)²/Expected, with df = categories − 1 for goodness of fit and (rows − 1)(columns − 1) for homogeneity or independence.
The p-value is the probability of getting a test statistic as extreme or more extreme than yours, assuming the null hypothesis is true.
Test statistic formulas aren't printed as labeled formulas on the AP exam sheet, but you can rebuild them from the general formula and the standard error formulas that are provided.
Always report degrees of freedom alongside t and chi-square test statistics, since the p-value depends on both.
It's a standardized value calculated from sample data that measures how far your observed result is from what the null hypothesis predicts. For z- and t-tests it follows (statistic − parameter)/standard error; for chi-square tests it's Σ(Observed − Expected)²/Expected.
No. The test statistic is a distance measure (like t = 2.31 means your sample mean is 2.31 standard errors above the null value), while the p-value is the probability of getting a test statistic that extreme if the null were true. You compute the test statistic first, then use it to find the p-value.
Not exactly. The CED states the formulas don't appear explicitly on the formula sheet, but you can construct them from the general pattern (statistic − parameter)/standard error using the standard error formulas the sheet does provide. The chi-square formula is one worth knowing cold.
Use z for proportions (categorical data, one or two groups), t for means (quantitative data, since you estimate σ with s), and chi-square when categorical data has more than two categories or you're comparing distributions in a two-way table. Picking the right one is the focus of Topic 8.7.
Yes. A test statistic farther from zero (or a larger chi-square value) means your data is more inconsistent with the null hypothesis, which pushes the p-value down. For example, χ² = 9.72 with 3 degrees of freedom gives a p-value below 0.05, so you'd reject H₀ at that level.