How to Perform a Paired T-Test (Step by Step)

A paired t-test compares two related measurements on the same subjects — like before and after a treatment, left vs. right hand, or two different methods on the same sample. Here’s exactly how to do it.

When to Use a Paired T-Test

Use a paired t-test when:

• You measure the same subjects under two conditions
• You have before and after data (pre-test / post-test)
• Subjects are matched in pairs (e.g., twins, matched controls)
• Your data is approximately normally distributed (or n > 30)

Don’t use it when: You’re comparing two independent groups with different subjects — use an independent two-sample t-test instead.

The 6 Steps

Step 1: State Your Hypotheses

• H₀ (null): The mean difference is zero (no effect)
• H₁ (alternative): The mean difference is not zero (there is an effect)

For a two-tailed test: H₁: μ_d ≠ 0 For a one-tailed test: H₁: μ_d > 0 or H₁: μ_d < 0

Step 2: Calculate the Differences

For each pair, compute d = After − Before (or whichever direction makes sense for your question).

Example: A new teaching method is tested on 8 students.

Student	Before	After	d (After − Before)
1	72	78	+6
2	65	70	+5
3	80	85	+5
4	68	72	+4
5	75	82	+7
6	82	84	+2
7	70	76	+6
8	77	80	+3

Step 3: Calculate the Mean and SD of Differences

• Mean difference (d̄): (6+5+5+4+7+2+6+3) / 8 = 38 / 8 = 4.75
• SD of differences (s_d): 1.669 (use the Standard Deviation Calculator)

Step 4: Compute the T-Statistic

$t = \frac{\bar{d}}{s_d / \sqrt{n}} = \frac{4.75}{1.669 / \sqrt{8}} = \frac{4.75}{0.590} = 8.05$

Step 5: Find the Critical Value or P-Value

• Degrees of freedom: df = n − 1 = 8 − 1 = 7
• From the t-table at α = 0.05 (two-tailed), df = 7: t-critical = 2.365
• Since |8.05| > 2.365, we reject H₀

Step 6: State Your Conclusion

“The new teaching method produced a statistically significant improvement in test scores (t(7) = 8.05, p < 0.001). Students scored an average of 4.75 points higher after the intervention.”

Checking Assumptions

Before trusting your results, verify:

1. Paired data: Each observation in one group corresponds to exactly one in the other
2. Continuous data: Differences should be measured on a continuous scale
3. Approximately normal differences: Plot the differences — they should be roughly symmetric. With n > 30, this is less critical due to the Central Limit Theorem
4. No extreme outliers: Outliers in the differences can distort results

Effect Size: Cohen’s d for Paired Data

Report effect size alongside your p-value:

$d = \frac{\bar{d}}{s_d} = \frac{4.75}{1.669} = 2.85$

Cohen’s d	Interpretation
0.2	Small effect
0.5	Medium effect
0.8	Large effect

Our d = 2.85 indicates a very large effect — the teaching method made a substantial difference.

Common Mistakes

1. Using an independent t-test when data is paired — this loses power and may miss real effects
2. Ignoring the direction of subtraction — be consistent (always After − Before, or Treatment − Control)
3. Reporting only the p-value — always include the mean difference, confidence interval, and effect size
4. Small sample sizes without checking normality — with n < 15, plot your differences to check

Try It Yourself

Enter your paired data directly into our T-Test Calculator — select “Paired” mode, enter both groups, and get instant results with step-by-step workings.

Need to look up a critical t-value? Use the t-Table.

Related Reading

• Hypothesis Testing: A Beginner’s Complete Guide
• When to Use Z-Test vs T-Test vs Chi-Square
• How to Choose the Right Statistical Test
• Degrees of Freedom Reference