t critical value, df = 21, 99% confidence
For 21 degrees of freedom the two-tailed t critical value at 99% confidence is 2.831. For a one-tailed test with α = 0.01 the critical value is 2.518. The matching z value is 2.576.
How to work it out
- 1
Degrees of freedom: for a single sample of size n they are n − 1. Here df = 21, so n = 22.
- 2
The confidence level 99% leaves α = 0.01 in the tails: half of it in each tail for a two-tailed interval.
- 3
Read the t-table at row df = 21 and the column for 0.005 in one tail.
- 4
Use it in the confidence interval for a mean when σ is unknown. For example, with x̄ = 50, s = 10 and n = 22:
- 5
The t value is larger than z = 2.576 because a small sample estimates the spread poorly; the difference is 9.9%, and it shrinks as df grows.
LaTeX source of the formulas on this page
\[ t^{*}_{21,\,99\%} = 2.831 \]
\[ df = n - 1 = 21 \]
\[ \frac{\alpha}{2} = 0.005 \]
\[ t^{*} = 2.831 \]
\[ 50 \pm 2.831\cdot\frac{10}{\sqrt{22}} = 50 \pm 6.036 = (43.964,\ 56.036) \]t critical values for df = 21
| Confidence | Two-tailed t* | One-tailed t (α = 1 − C) | z* |
|---|---|---|---|
| 80% | 1.323 | 0.859 | 1.282 |
| 90% | 1.721 | 1.323 | 1.645 |
| 95% | 2.080 | 1.721 | 1.960 |
| 98% | 2.518 | 2.189 | 2.326 |
| 99% | 2.831 | 2.518 | 2.576 |
t* at 99% confidence for other degrees of freedom
| df | t* |
|---|---|
| 1 | 63.657 |
| 2 | 9.925 |
| 5 | 4.032 |
| 10 | 3.169 |
| 20 | 2.845 |
| 30 | 2.750 |
| 60 | 2.660 |
| 120 | 2.617 |
Questions
What is the t critical value for df = 21 at 99% confidence?
2.831 (two-tailed).
What is the one-tailed critical value for df = 21 and α = 0.01?
2.518.
Why is t larger than z?
With few degrees of freedom the sample standard deviation is an unreliable estimate, so the t distribution has heavier tails and needs a wider interval. At df = 21 the t value is 9.9% larger than z = 2.576.
How many degrees of freedom does a sample of 22 have?
n − 1 = 21 for a one-sample t interval or test.