Chi-Square Table

Critical values of the chi-square distribution. Each column heading is the area to the right of the value, which is the α you use in a chi-square test. Enter your df and α above to highlight the right cell and get the exact value.

How to read the chi-square table

  1. Find your degrees of freedom. For a goodness-of-fit test, df = number of categories − 1. For a test of independence, df = (rows − 1) × (columns − 1).
  2. Pick the column for your α, usually 0.05.
  3. If your χ² statistic is larger than the table value, reject H₀.

Chi-square tests only use the right tail, because a large χ² means the observed counts are far from the expected counts. Small values never count against H₀.

A worked example

A shop records whether customers bought a product (yes or no) across three store layouts. The 3 × 2 table gives χ² = 6.80.

  • df = (3 − 1) × (2 − 1) = 2.
  • Row 2, column 0.05 gives 5.991. Show this in the table.
  • 6.80 is larger than 5.991, so the layout and buying are related at the 5% level.
  • Moving right along row 2, 6.80 sits between 5.991 (0.05) and 7.378 (0.025), so 0.025 < p < 0.05. The exact p-value is 0.033.

To run the whole test from your counts, including expected counts and Cramér’s V, use the chi-square test calculator.

What the left-hand columns are for

The columns from 0.995 to 0.90 are the lower tail. You need them for a confidence interval for a variance. With n = 10 and sample variance s² = 4, the 95% interval uses df = 9 and the values 19.023 (column 0.025) and 2.700 (column 0.975):

(9 × 4) / 19.023 = 1.89   to   (9 × 4) / 2.700 = 13.33

So the population variance is somewhere between 1.89 and 13.33. The interval is wide and lopsided, which is normal for variances from small samples. The variance confidence interval calculator does this for you.

Common mistakes

  • Using n − 1 for a test of independence. The df depends on the table’s shape, not the sample size. A 3 × 4 table always has df = 6, whether it holds 50 people or 5,000.
  • Reading a left-tail column for a test. For a chi-square test, use the column that matches α (0.05, 0.01). The 0.95 column is for intervals, not tests.
  • Trusting the test with tiny expected counts. If more than a fifth of the expected counts are below 5, the table values are not reliable. For a 2 × 2 table, use Fisher’s exact test.

Check yourself

A die is rolled 120 times to check if it is fair. What is the critical value at α = 0.05?

Six faces means df = 6 − 1 = 5. Row 5, column 0.05: 11.070.

A 2 × 2 test of independence gives χ² = 3.20. Significant at 0.05?

df = 1 and the critical value is 3.841. Since 3.20 is smaller, the result is not significant.