Paired t-Test: Comparing Dependent Samples
What you'll learn: How to test whether the mean difference between paired observations is significantly different from zero.
The Core Idea
A paired t-test is used when you have dependent samples—two measurements on the same subjects or matched pairs. Think of it as comparing "before" and "after" measurements: weight before and after a diet, test scores before and after tutoring, or blood pressure readings from the same patients under two different conditions.
Instead of comparing two separate groups, you calculate the difference for each pair, then test whether the average difference is significantly different from zero (or some other hypothesized value).
Why It's Different
Unlike a two-sample t-test (which compares independent groups), the paired t-test accounts for natural variability between subjects. By looking at differences within each pair, you control for individual differences and gain more statistical power.
The Process
- Calculate differences: For each pair (subject), compute d = measurement₁ - measurement₂
- Compute mean difference: d̄ (average of all differences)
- Calculate standard deviation of differences: sᴅ
- Compute the t-statistic: t = d̄ / (sᴅ/√n), where n is the number of pairs
- Compare to t-distribution with (n-1) degrees of freedom
The null hypothesis typically states H₀: μᴅ = 0 (no mean difference), and you reject it if |t| exceeds the critical value for your chosen significance level.
Real-World Analogy
Imagine testing a new teaching method by measuring students' scores before and after. Each student serves as their own control, eliminating variation due to individual ability levels. You're asking: "Does the method produce a consistent change across students?"
Key Takeaway: The paired t-test converts paired measurements into a single set of differences, then performs a one-sample t-test on those differences to determine if the mean change differs from zero.