Calculate the Chi-Square (χ²) test statistic and degrees of freedom from observed and expected frequencies.
📖 Pearson's Chi-Square Goodness-of-Fit FormulaFill in the fields on the left with your information, then press Calculate to see your result instantly. No sign-up required, and no data is sent anywhere — everything is calculated right in your browser.
The Chi-Square test is a statistical test that compares the actually "observed" values against the theoretically "expected" values to see how much they agree or differ, and it's widely used in the "goodness of fit" test to check whether actual data matches an expected distribution. The χ² value is calculated by summing, for each data category, the squared difference between each observed and expected value divided by the expected value itself. The closer the χ² value is to zero, the greater the agreement between the observed and expected data, and the larger the value, the greater the difference. This calculator also shows the "degrees of freedom," which is the number of categories minus one, needed to statistically interpret the result by comparing it against a chi-square distribution table. A large χ² value on its own does not automatically prove a meaningful difference — its significance always needs to be interpreted in the context of the degrees of freedom and a chosen significance level, typically using a chi-square distribution table or equivalent software function.
The Chi-Square goodness-of-fit test answers a very specific and commonly useful question: given a set of observed counts across several categories, are the differences from a theoretically expected set of counts large enough to be statistically meaningful, or small enough to be reasonably attributed to ordinary random variation?
The calculation itself has an intuitive structure once broken into its parts. For each category, the difference between what was actually observed and what was theoretically expected is calculated, then squared (to eliminate the sign and penalize larger gaps more heavily), then divided by the expected value for that category (to scale the discrepancy relative to how large a difference is reasonable to expect in that category in the first place). Summing this ratio across every category produces the final χ² statistic.
A textbook example makes the logic concrete: rolling a fair six-sided die 60 times should, in theory, produce each face roughly 10 times. If the actual observed counts are close to 10 for every face, the χ² value will be small, supporting the hypothesis that the die is fair. If one face came up 25 times and another only twice, the χ² value will be large, suggesting the die may not actually be fair — though 'suggesting' is the operative word, since some deviation from the theoretical expectation is normal even with a genuinely fair die due to ordinary chance.
Degrees of freedom — one less than the number of categories — matter because the same χ² value means different things depending on how many categories produced it. A χ² value of 10 is a much stronger signal of a genuine discrepancy with only 2 categories (2 degrees of freedom) than with 10 categories (9 degrees of freedom), since more categories naturally accumulate more random variation even under a completely true null hypothesis. Comparing the calculated χ² value against a chi-square distribution table, using the correct degrees of freedom, is how a formal statistical conclusion (statistically significant or not, at a chosen threshold like 5%) is actually reached.
Beyond dice and coins, the Chi-Square test is used throughout genetics (testing whether observed offspring trait ratios match theoretical Mendelian predictions), market research (testing whether customer preferences differ meaningfully across demographic groups), and quality control (testing whether defect rates differ meaningfully across production batches or shifts).
A low value suggests the observed data closely matches the expected distribution, with little deviation.
Because the formula divides by the expected value, so a zero would make the calculation mathematically undefined.
No, the statistic must be compared against a Chi-Square distribution table using the degrees of freedom and a chosen significance level to determine statistical significance.