What is a Critical Value Calculator?
A critical value is the cut-off a test statistic must reach for a result to be statistically significant at significance level α. This critical value calculator finds it for the z, t, chi-square (χ²), F and Pearson r distributions, for left-, right- and two-tailed tests. At α = 0.05, two-tailed, the z critical value is ±1.96; t, χ², F and r values also depend on the degrees of freedom or sample size.
What the Critical Value Calculator does
Pick the Distribution your test statistic follows, enter its degrees of freedom (or the sample size for a correlation), give α or a confidence level, and choose the tail. The calculator returns the critical value or values, the rule for rejecting H₀, a curve with the rejection region shaded, and a table showing how the cut-off moves across common α levels. Chi-square and F default to a right-tailed test because that is how goodness-of-fit, independence, ANOVA and regression tests use them.
To calculate a critical value by hand you invert the cumulative distribution function (CDF) of the test statistic at 1 − α, α or 1 − α/2. Only the normal case is well known (1.645, 1.96, 2.576); the inverse t, χ² and F functions have no closed form, which is why printed tables exist. The calculator solves them numerically, so it works for any α and for degrees of freedom, including decimals, that a table skips.
This page is the general-purpose tool: one form for all five distributions, useful when you are moving between tests or are not yet sure which table you need. Each distribution also has a dedicated page with more depth: the Z Critical Value Calculator, T Critical Value Calculator, Chi-Square Critical Value Calculator, F Critical Value Calculator and Pearson r Critical Value Calculator. If you already have a test statistic and want its exact probability instead of a cut-off, use the P-Value Calculator.
How to use the Critical Value Calculator
- Choose the Distribution: z, t, χ², F or r. The table further down this page matches common tests to distributions.
- Enter the Degrees of freedom (df) for t or χ², the Numerator df and Denominator df for F, or the Sample size (n pairs) for r. The z distribution needs none.
- Under Enter the, choose Significance level α or Confidence level, and type the value (0.05, or 95).
- Choose the Tail: two-tailed for “not equal to”, or left or right for a directional alternative.
- Read the critical value and the decision rule, then compare your own test statistic with it.
Critical Value formula
A critical value is a quantile: the point on the test statistic’s null distribution that cuts off a tail of area α (or α/2 in each tail). Q(p) below is the inverse CDF of the chosen distribution.
| Symbol | Meaning |
|---|---|
| α | significance level, the probability of rejecting a true H₀; α = 1 − confidence level |
| Q(p) | quantile function: Φ⁻¹ for z, and the inverse t, χ² or F CDF with the stated degrees of freedom |
| c, c₁, c₂ | critical values bounding the rejection region |
| n | number of (x, y) pairs for a correlation |
The chi-square and F distributions are not symmetric, so a two-tailed test has two unrelated critical values and no “±” shortcut.
For a two-sided confidence interval, use the two-tailed critical value at α = 1 − confidence level.
Worked example
Worked example: the same α on five distributions
Problem. At α = 0.05, find the critical value for (a) a two-tailed z-test, (b) a two-tailed t-test with n = 25, (c) a chi-square test of independence on a 3 × 3 table, (d) a one-way ANOVA with 4 groups and 40 observations, and (e) a two-tailed correlation test with 25 pairs.
- 01(a) z, two-tailed: Φ⁻¹(0.975) = 1.96, so reject H₀ if |z| ≥ 1.96.
- 02(b) t, two-tailed, df = 25 − 1 = 24: T⁻¹(0.975; 24) = 2.0639, so reject if |t| ≥ 2.0639.
- 03(c) χ², right-tailed, df = (3 − 1)(3 − 1) = 4: the 0.95 quantile is 9.4877, so reject if χ² ≥ 9.4877.
- 04(d) F, right-tailed, df₁ = 4 − 1 = 3 and df₂ = 40 − 4 = 36: the 0.95 quantile is 2.8663, so reject if F ≥ 2.8663.
- 05(e) r, two-tailed, df = 25 − 2 = 23: t = 2.0687, and r = 2.0687 / √(2.0687² + 23) = 0.3961, so reject if |r| ≥ 0.3961.
Answer. The critical values are ±1.96 (z), ±2.0639 (t, 24 df), 9.4877 (χ², 4 df), 2.8663 (F, 3 and 36 df) and ±0.3961 (r, n = 25). Same α, five different cut-offs, because each statistic has its own sampling distribution.
Which distribution does my test use: z, t, χ², F or r?
The critical value always comes from the distribution the test statistic follows when H₀ is true. Find your test in the table, then use this page or follow the link for the dedicated calculator and its printable table.
| Test or interval | Distribution | Degrees of freedom | Usual tail | Dedicated page |
|---|---|---|---|---|
| Mean with σ known; one or two proportions | z | none | Two, left or right | Z Critical Value Calculator |
| Mean with σ unknown; paired and two-sample t-tests; regression slope | t | n − 1, n₁ + n₂ − 2, or n − k − 1 | Two, left or right | T Critical Value Calculator |
| Goodness of fit; independence in a contingency table | χ² | k − 1, or (r − 1)(c − 1) | Right | Chi-Square Critical Value Calculator |
| One variance or standard deviation | χ² | n − 1 | Two, left or right | Chi-Square Critical Value Calculator |
| One-way ANOVA; overall regression F-test | F | k − 1 and N − k | Right | F Critical Value Calculator |
| Ratio of two variances | F | n₁ − 1 and n₂ − 1 | Two or right | F Critical Value Calculator |
| Pearson correlation different from zero | r (via t) | n − 2 | Two, left or right | Pearson r Critical Value Calculator |
Critical value table: common z, t, χ², F and r values
These are the values most often looked up, at the three usual significance levels. Two-tailed values for the symmetric z, t and r distributions are shown with ±; chi-square and F are right-tailed, the way their tests use them. For every degree of freedom, use the full z table, t-table, chi-square table, F-table or r critical value table, or enter your own values above.
| Distribution | Tail | α = 0.10 | α = 0.05 | α = 0.01 |
|---|---|---|---|---|
| z | Two-tailed | ±1.6449 | ±1.9600 | ±2.5758 |
| z | One-tailed (right) | 1.2816 | 1.6449 | 2.3263 |
| t, df = 10 | Two-tailed | ±1.8125 | ±2.2281 | ±3.1693 |
| t, df = 20 | Two-tailed | ±1.7247 | ±2.0860 | ±2.8453 |
| t, df = 30 | Two-tailed | ±1.6973 | ±2.0423 | ±2.7500 |
| t, df = 60 | Two-tailed | ±1.6706 | ±2.0003 | ±2.6603 |
| χ², df = 1 | Right | 2.7055 | 3.8415 | 6.6349 |
| χ², df = 2 | Right | 4.6052 | 5.9915 | 9.2103 |
| χ², df = 4 | Right | 7.7794 | 9.4877 | 13.2767 |
| F, df₁ = 2, df₂ = 30 | Right | 2.4887 | 3.3158 | 5.3903 |
| F, df₁ = 3, df₂ = 36 | Right | 2.2426 | 2.8663 | 4.3771 |
| r, n = 10 pairs | Two-tailed | ±0.5494 | ±0.6319 | ±0.7646 |
| r, n = 30 pairs | Two-tailed | ±0.3061 | ±0.3610 | ±0.4629 |
One-tailed vs two-tailed critical values
A two-tailed test asks whether the parameter differs from the null value in either direction, so α is split equally: α/2 in each tail. A one-tailed test asks about one direction only, so all of α sits in that tail and the cut-off is closer to the centre. At α = 0.05 the z critical value is ±1.96 two-tailed but 1.645 one-tailed; with 24 df the t values are ±2.0639 and 1.7109.
For the symmetric z, t and r distributions, the one-tailed critical value at α equals the two-tailed value at 2α. Chi-square and F are skewed, so a two-tailed test has two unrelated cut-offs: for χ² with 4 df at α = 0.05 they are 0.4844 and 11.1433, not a ± pair.
- Two-tailed: H₁ uses ≠. Also the right choice for a two-sided confidence interval.
- Right-tailed: H₁ uses >, and every χ² goodness-of-fit, independence and ANOVA F-test.
- Left-tailed: H₁ uses <, for example a mean or variance claimed to be below a target.
- Choose the tail before looking at the data. Switching from two-tailed to one-tailed after seeing the direction doubles the real Type I error rate.
Critical value vs test statistic
The test statistic (z, t, χ², F or r) is computed from your sample and changes with the data. The critical value is fixed before you look at the data: it depends only on the distribution, the degrees of freedom, α and the tail. The decision comes from comparing the two.
For example, a two-tailed one-sample t-test with n = 25 gives t = 2.45. The critical value at α = 0.05 with 24 df is ±2.0639. Because |2.45| ≥ 2.0639, the statistic is in the rejection region and H₀ is rejected. Had the statistic been t = 1.80, it would fall short of the cut-off and you would fail to reject H₀.
Critical value approach versus p-value approach
The two approaches always give the same decision. With the critical value approach you fix α first, look up the cut-off, and reject H₀ if the test statistic lands in the rejection region. With the p-value approach you compute the probability of a statistic at least as extreme as yours and reject H₀ if that probability is at most α.
A statistic exactly equal to the critical value has a p-value exactly equal to α. The critical value is quicker when you only need a yes-or-no decision or are working from printed tables; the p-value says how strong the evidence is, which is why journals ask for it. The P-Value Calculator handles that side.
Critical values in Excel, R and Python
Most functions take the cumulative (left-tail) probability, so a right-tailed critical value at α is the quantile at 1 − α. Excel also offers right-tail and two-tail versions that take α directly.
| Distribution | Excel / Google Sheets | R | Python (SciPy) |
|---|---|---|---|
| z, two-tailed | NORM.S.INV(1 − α/2) | qnorm(1 − α/2) | norm.ppf(1 − α/2) |
| t, two-tailed | T.INV.2T(α, df) | qt(1 − α/2, df) | t.ppf(1 − α/2, df) |
| χ², right-tailed | CHISQ.INV.RT(α, df) | qchisq(1 − α, df) | chi2.ppf(1 − α, df) |
| F, right-tailed | F.INV.RT(α, df1, df2) | qf(1 − α, df1, df2) | f.ppf(1 − α, df1, df2) |
| r, two-tailed | from t: t / SQRT(t^2 + n − 2) | from t: t / sqrt(t^2 + n − 2) | from t: t / sqrt(t**2 + n − 2) |
How to find a critical value on a TI-83 or TI-84 calculator
The inverse functions are in the DISTR menu (2nd, VARS). Like Excel’s NORM.S.INV, they take the area to the left of the cut-off, so a right-tailed critical value uses 1 − α and a two-tailed one uses 1 − α/2.
- z: invNorm(1 − α/2, 0, 1) for a two-tailed test, for example invNorm(0.975, 0, 1) = 1.96.
- t: invT(1 − α/2, df), for example invT(0.975, 24) = 2.0639. invT is on TI-84 models with operating system 2.30 or later, not on the TI-83.
- χ² and F: there is no inverse function on the TI-83 or TI-84, so use a printed table or this calculator.
- r: find the critical t with n − 2 df, then r = t / √(t² + n − 2).
Common use cases
- Setting up a hypothesis test by hand: find the rejection region before computing the test statistic.
- Building confidence intervals: the two-tailed critical value is the multiplier in estimate ± critical value × standard error.
- Checking a printed table: get the exact value for degrees of freedom that a table skips or rounds.
- Homework and exams across several chapters: switch between z, t, χ², F and r without changing tools.
- Planning: see how much the cut-off rises if you tighten α from 0.05 to 0.01.
- Learning to read printed tables: the guide to statistical tables and critical values shows how to read the z, t, χ², F and r tables by hand, with worked lookups.
Assumptions and limitations
- A critical value is only as valid as the test behind it. If the test’s assumptions fail (non-normal small samples for t, small expected counts for χ², unequal variances for ANOVA), the nominal α no longer applies.
- Choosing the tail after looking at the data inflates the Type I error rate. Decide between one- and two-tailed before you collect data.
- The calculator covers z, t, χ², F and r. Other statistics, such as the studentized range for Tukey’s test or the Mann–Whitney U, have their own tables.
- A critical value gives a yes-or-no decision at one α. It does not measure the size of the effect; report an estimate with a confidence interval, or an effect size, alongside it.
- Very small significance levels (below 10⁻¹⁰) are outside the range this calculator evaluates.
Frequently asked questions
What is a critical value in statistics?
How do I calculate the critical value?
What is the critical value for a 95% confidence level?
Should I use a z or a t critical value?
Why are chi-square and F critical values usually right-tailed?
What is a critical z value?
What is the difference between a critical value and a p-value?
Cite this page
Statculator. (2026, November 14). Critical Value Calculator. Retrieved November 14, 2026, from https://statculator.com/p-values-critical-values/critical-value-calculator/