Overview
A t statistic measures how many standard errors a sample result is from what the null hypothesis predicts. This turns a t value into a p-value and a decision.
Enter the t statistic, the degrees of freedom, distribution code 2, the number of tails and alpha. Calculate t first, for example as (x̄ − μ₀) ÷ (s ÷ √n).
Worked examples
| Inputs | Result |
|---|---|
| Test Statistic2.15Degrees of Freedom28DistributiontTailsTwo-tailedAlpha (%)5 | Distribution: tTest statistic: 2.15Degrees of freedom: 28p-value: 0.04Reject H0 at alpha=5% |
Show 2 more examples
| Test Statistic3.1Degrees of Freedom40DistributiontTailsTwo-tailedAlpha (%)5 | Distribution: tTest statistic: 3.1Degrees of freedom: 40p-value: 0.00354Reject H0 at alpha=5% |
| Test Statistic1.5Degrees of Freedom60DistributiontTailsOne-tailedAlpha (%)5 | Distribution: tTest statistic: 1.5Degrees of freedom: 60p-value: 0.07Fail to reject H0 at alpha=5% |
How it works
The p-value comes from the t distribution with your degrees of freedom. With fewer degrees of freedom the tails are heavier, so the same t gives a larger p-value.
The formula
| x̄ | Sample mean |
| μ₀ | Value under the null hypothesis |
| s | Sample standard deviation |
Practical tips
- t = 2.15 with 28 df gives a two-tailed p of about 0.04, which is significant at 5%.
- You can check it in Excel with =T.DIST.2T(ABS(t), df).
- Degrees of freedom for a one-sample t-test are n − 1.
Frequently asked questions
What is a large t value?
Beyond about ±2 is often significant at 5% for moderate samples.
Why are the degrees of freedom needed?
The t distribution depends on them. Smaller samples have heavier tails, so they need a larger t to reach significance.
One or two tails?
Use two unless you decided in advance to test only one direction.