Skip to content
AP Statistics · Inference for Quantitative Data: Means

Two sample t test

Two sample t test appears in 26 reviewed AP Statistics questions and sits in Inference for Quantitative Data: Means. Below are worked examples showing exactly how the exam tests it, with the answer marked and every distractor explained.

How Two sample t test is tested

Example 1Medium
A two-sample t-test examines if mean test scores differ between students who study in groups versus alone. The study is based on a convenience sample from one university. Even if the test is statistically significant, what is the primary limitation?
  1. AStatistical significance guarantees the difference is educationally meaningful.
  2. BThe t-test assumes equal variances, which may not hold.
  3. CThe two-sample t-test cannot detect mean differences.
  4. DResults from one university may not generalize.Correct

Why D is right

While statistical significance indicates a real effect in the sample, a convenience sample from one university may not represent all students. Students at one institution differ in many ways from the broader student population. The results may not generalize (lack external validity).
Why the other 3 choices miss
A
Statistical significance does not guarantee practical or educational significance. A small but statistically significant difference may not matter in practice — but this choice also doesn't address the primary generalizability concern.
B
Welch's two-sample t-test (standard in AP Statistics) does not require equal variances. Equal variance is not a necessary assumption.
C
Two-sample t-tests are specifically designed to detect mean differences. This claim is false.
Example 2Medium
A two-sample t-test compares the mean test scores of students taught by Method A (n1 = 25, x-bar1 = 78, s1 = 10) and Method B (n2 = 30, x-bar2 = 83, s2 = 12). The test has H0: mu1 - mu2 = 0 and Ha: mu1 - mu2 ≠ 0. The standard error of the difference in sample means is approximately
  1. A(10/25) + (12/30) = 0.40 + 0.40 = 0.80
  2. Bsqrt(10^2 + 12^2) = sqrt(244) ≈ 15.62
  3. Csqrt((10 + 12)^2 / (25 + 30)) = sqrt(484/55) ≈ 2.97
  4. Dsqrt(10^2/25 + 12^2/30) = sqrt(4 + 4.8) ≈ 2.97Correct

Why D is right

The standard error of the difference in independent sample means is SE = sqrt(s1^2/n1 + s2^2/n2) = sqrt(100/25 + 144/30) = sqrt(4 + 4.8) = sqrt(8.8) ≈ 2.97. Each sample contributes its own variance divided by its sample size.
Why the other 3 choices miss
A
Incorrect. This divides the standard deviations (not variances) by the sample sizes and adds them. The correct formula uses variances (squared standard deviations) divided by sample sizes.
B
Incorrect. This computes sqrt(s1^2 + s2^2), which ignores the sample sizes entirely. The standard error must account for how sample size reduces variability.
C
Incorrect. This incorrectly pools the standard deviations by adding them before squaring and dividing by the combined sample size. The correct formula keeps the samples separate: s1^2/n1 + s2^2/n2.

Studied alongside

Other concepts from Inference for Quantitative Data: Means.