Part of the Maths & Statistics suite · 235 calculators

Paired T-Test Calculator

Test whether a before-and-after change is real — the paired t-test, which works on the differences and is far more powerful than comparing two groups.

When each observation is measured twice, the pairing removes between-subject variation entirely.

Comma-separated
Same count, same order
Results update as you type
Results
t statistic
8.282512
p-value (two-tailed)
Degrees of freedom
Decision
Mean difference
SD of the differences
Standard error
CI — lower
CI — upper
Cohen's dz
Reviewed September 2026. Pure mathematics: the result does not depend on where you are. Terminology follows the national curriculum (maths, brackets, decimal point).
No account required · Google Analytics off unless allowedCalculator arithmetic runs in your browserResults update as you type
All calculations run 100% in your browser. The calculator code does not submit your figures to GlobalCalc to obtain a result.
About paired t-test

How the paired t-test calculator works

When each observation is measured twice, the pairing removes between-subject variation entirely. Compute each difference, then run a one-sample t-test on those differences against zero: t = d̄ ÷ (s_d / √n).

That is why paired designs are so much more sensitive. The same data analysed as two independent groups often fails to reach significance, because it is fighting variation the pairing had already eliminated.

Formula: t = d̄ / (s_d / √n), df = n − 1

Worked examples

Inputst statisticNote
Eight before-and-after pairs8.282512a clear drop, strongly significant
No change0the changes cancel
A small consistent shift5small but consistent — highly significant

Frequently asked questions

When is a paired test right?

When the same subjects, items or matched pairs are measured twice. Before-and-after, left-and-right, treatment-and-control on matched pairs.

Why is it more powerful?

Because it removes all the variation between subjects. Only the change matters, so a small consistent effect stands out clearly.

What if I analysed it as two independent groups?

You would throw away the pairing and usually fail to detect a real effect. It is one of the most common analysis errors.

What does Cohen's dz mean?

The mean change divided by the standard deviation of the changes — the effect size for a paired design. It is not comparable to the two-group d.

What if the differences are badly skewed?

Use the Wilcoxon signed-rank test, or the sign test, which make no distributional assumption.

Where these figures come from

Last checked: September 2026. Formulas are fixed by mathematics and do not change with tax years or regulations.