Skip to content
Omnikalc

Paired t-test calculator: is the mean difference between paired measurements real?

Enter the mean and standard deviation of the paired differences and the number of pairs to get the t statistic, the p-value, the confidence interval and the effect size.

Updated on
Your inputs

Value out of range

Value out of range

Value out of range

Results

t statistic

2.305

Degrees of freedom
23
p-value (two-sided)
0.0305
Significant at the chosen level
Confidence interval of the mean difference: lower
0.246
Confidence interval of the mean difference: upper
4.554
Effect size dz (mean difference ÷ SD of differences)
0.471

A paired t-test checks whether the average change between two linked measurements differs from zero. The paired t-test calculator takes the mean of the differences, their standard deviation, the number of pairs and a significance level, and gives the t statistic, the degrees of freedom, the p-value, the confidence interval of the mean difference and the effect size dz.

How the calculation works

  • For each pair, take the difference (after minus before). The test works on these differences, with their mean d̄ and standard deviation s.
  • The standard error is s ÷ √n, and the t statistic is d̄ ÷ (s ÷ √n), with n − 1 degrees of freedom.
  • The p-value is the two-sided probability of a t at least this far from zero. The confidence interval is d̄ ± t(1 − α ÷ 2; n − 1) × s ÷ √n.
  • The effect size dz is d̄ ÷ s. A small effect is around 0.2, a medium one 0.5 and a large one 0.8.

Example

Weights are measured before and after a program for 24 people. The mean difference is 2.4 and the standard deviation of the differences is 5.1. The standard error is 1.041, so t = 2.305 with 23 degrees of freedom and the p-value is 0.0305. The 95 % interval of the mean difference is 0.246 to 4.554, which excludes 0, and dz is 0.471: a significant, small to medium effect.

Why it matters

Before-and-after designs are common in medicine, education and testing. When you only have the summary of the differences from a paper or a spreadsheet, this calculator gives the test result without needing the raw data.

Good to know

The test needs the differences, not the two group means, so compute the difference for each pair first. It assumes approximate normality of the differences, which matters for samples under about 30. The p-value tells whether an effect is likely real, not whether it is large, so look at the effect size and the interval.

The t-test calculator compares two independent groups, and the one-sample t-test calculator compares a mean with a fixed value.

Frequently asked questions

When should I use a paired t-test?

Use it when each observation in one sample is linked to one in the other, as with before and after measurements on the same people, matched twins or two methods applied to the same items. The pairing removes the variation between subjects, which makes the test more sensitive.

What are the assumptions?

The differences should be roughly normally distributed, especially for small samples, and the pairs should be independent of each other. It does not require the original measurements to be normal. If the differences are strongly skewed, the Wilcoxon signed-rank test is an alternative.

Why use the differences and not two separate samples?

Treating paired data as two independent samples ignores the link between them and loses power. A paired test works with a single sample of differences, so the person-to-person variation drops out and only the change remains.

What does the confidence interval tell me?

It gives the range of mean differences that are consistent with the data. If it excludes zero, the test is significant at the matching level. Its width also shows how precisely the change is estimated, which a p-value alone does not.

Was this calculator helpful?

Thanks for your feedback!

Results are estimates provided for information purposes. They do not replace professional advice.