Overview

A critical value is the cutoff a test statistic must pass to be significant. This finds it for z, t or chi-square tests at your confidence level.

Enter the confidence level, a distribution code (1 z, 2 t, 3 chi-square), the degrees of freedom and one or two tails.

Worked examples

InputsResult
Confidence Level (%)95Distributionz (normal)Degrees of Freedom30TailsTwo-tailedCritical z value: 1.96Confidence: 95%Tails: two-taileddf: 30
Show 3 more examples
Confidence Level (%)99Distributionz (normal)Degrees of Freedom30TailsTwo-tailedCritical z value: 2.58Confidence: 99%Tails: two-taileddf: 30
Confidence Level (%)95DistributiontDegrees of Freedom10TailsTwo-tailedCritical t value: 2.23Confidence: 95%Tails: two-taileddf: 10
Confidence Level (%)95DistributionChi-squareDegrees of Freedom5TailsOne-tailedCritical chi-square value: 11.07Confidence: 95%Tails: one-taileddf: 5

How it works

The z value comes from the inverse normal curve. The t and chi-square values are solved numerically from their exact distributions, so they match printed tables.

The formula

Two-tailed z = Φ⁻¹(1 − α ÷ 2)
α1 − confidence level

Practical tips

  • 95% two-tailed z is 1.96, and 99% is 2.576.
  • For t with 10 df at 95% two-tailed, the critical value is 2.23 (2.228 in tables).
  • Chi-square tests are usually one-tailed. At 95% with 5 df, the critical value is 11.07.

Frequently asked questions

One or two tails?

Two tails if a difference in either direction matters, one tail if only one direction does.

Why use t instead of z?

When the standard deviation is estimated from a small sample.

How can I check these?

Excel's T.INV.2T and CHISQ.INV.RT functions, or printed tables.