Choosing between the chi-square test and Fisher's exact test is one of the most common decisions researchers face when analyzing categorical data. The short answer: use Fisher's exact test when expected cell counts fall below 5 in any cell of your contingency table, and use the chi-square test when expected counts are adequate.
This guide walks you through every decision point.
Try our free Chi-Square Calculator to run either test instantly with automatic assumption checking.
Why Expected Cell Counts Drive the Decision
The chi-square test is an approximation. The key metric is the expected cell frequency in each cell of your contingency table.
The rule: if any expected cell count is less than 5, switch to Fisher's exact test. Calculate expected counts: for each cell, multiply its row total by its column total, divide by the grand total.
Fisher's Exact Test
Fisher's exact test calculates the exact probability of observing your data given the null hypothesis. It does not rely on an approximation, which is why it remains valid regardless of sample size. It is primarily designed for 2x2 tables; for larger tables, the Freeman-Halton test generalizes the logic.
Chi-Square Test Assumptions
Independence of observations, adequate expected frequencies (at least 5 in all cells), and categorical data.
Yates' continuity correction subtracts 0.5 from the absolute difference between observed and expected counts. Many statisticians now consider it overly conservative and recommend against it except when N is small.
Decision Table
| Situation | Recommended Test |
|---|---|
| 2x2 table, all expected counts above 5 | Chi-square |
| 2x2 table, any expected count below 5 | Fisher's exact test |
| Larger table, all expected counts above 5 | Chi-square |
| Larger table, any expected count below 5 | Freeman-Halton or Monte Carlo |
| N below 20 | Fisher's exact test |
| Paired categorical data | McNemar's test |
Effect Sizes: Cramer's V and Phi Coefficient
For 2x2 tables, use the phi coefficient (ranges 0 to 1). Benchmarks: small = 0.10, medium = 0.30, large = 0.50.
For larger tables, use Cramer's V, which scales phi by table dimensions.
Always report effect size alongside your p-value. A significant chi-square with Cramer's V of 0.05 suggests a trivially small association.
For diagnostic test accuracy data, see our Diagnostic Accuracy Calculator.
Applying This in Systematic Reviews
In systematic reviews, check whether original study authors chose the appropriate test given their sample sizes. Chi-square is also the backbone of Cochran's Q test for heterogeneity in meta-analysis.
For inter-rater agreement on categorical outcomes, the Kappa Calculator handles Cohen's kappa.
Common Mistakes
Checking total N instead of expected counts. Applying chi-square to paired data. Forgetting to report effect sizes. Assuming Fisher's is always more conservative (for large samples, both give similar p-values). Using Yates' correction reflexively.
Key Takeaways
- Use Fisher's exact test when any expected cell count is below 5 or when total N is below 20.
- Use the chi-square test for larger samples where all expected cell counts meet the threshold.
- For larger tables with small expected counts, use Freeman-Halton or Monte Carlo simulation.
- Always report phi (2x2 tables) or Cramer's V (larger tables) as your effect size measure.
- Paired categorical data requires McNemar's test, not chi-square or Fisher's.
- Statistical significance and effect size together tell the complete story.
FAQ
What is the main difference between chi-square and Fisher's exact test?
The chi-square test uses a mathematical approximation; Fisher's exact test calculates the exact probability. The chi-square approximation is reliable only when expected cell counts are large enough.
When should I use Fisher's exact test over chi-square?
When any cell has an expected count below 5, or when total N is below 20.
What is Yates' continuity correction?
Yates' correction adjusts the chi-square statistic by subtracting 0.5. It is typically applied to 2x2 tables when N is small but expected counts are technically above 5. Many modern statisticians recommend against routine use.
How do I calculate an effect size after a chi-square test?
For 2x2 tables: phi = sqrt(chi-square / N). For larger tables: Cramer's V = sqrt(chi-square / (N * min(rows-1, cols-1))).
Can I use chi-square for paired data?
No. Use McNemar's test for paired or matched categorical data.
How does this apply to systematic review data extraction?
Check whether original authors used the appropriate test given their sample sizes. Flag studies that applied chi-square with inadequate expected cell counts.
Need help with your systematic review or meta-analysis? Get a free quote from our team of PhD researchers.