Conducting Fisher's Exact Test
What you'll learn: How to apply Fisher's Exact Test to small 2×2 contingency tables and interpret the resulting p-values.
Why Conduct Fisher's Exact Test?
You learned the logic behind Fisher's Exact Test in the previous lesson—now it's time to actually carry it out. This test is used when your 2×2 contingency table has small expected frequencies (typically when any expected count is below 5), making the Chi-Squared test unreliable.
Setting Up Your Test
Start with your 2×2 contingency table (two categories for each of two variables). For example:
| Treatment Success | Treatment Failure | |
|---|---|---|
| Drug A | 8 | 2 |
| Drug B | 1 | 9 |
Next, define your hypotheses:
- Null (H₀): The two variables are independent (no association)
- Alternative (Hₐ): The variables are associated
One-Tailed vs Two-Tailed
- Two-tailed test: Tests if there's any association in either direction. The p-value sums probabilities of all tables as extreme or more extreme than observed (in both directions).
- One-tailed test: Tests if one specific group has higher success than expected. The p-value includes only tables showing that directional extreme.
Use two-tailed unless you had a specific directional hypothesis before collecting data.
Interpreting the P-Value
Fisher's Exact Test calculates the exact probability of observing your table (and more extreme ones) under independence. If p < α (usually 0.05), reject the null hypothesis—conclude there's evidence of association between the variables.
Key Takeaway: Fisher's Exact Test applies to small 2×2 tables by computing exact probabilities; use two-tailed p-values unless you had a pre-specified directional hypothesis.