Fisher's Exact Test is a hypothesis test for two categorical variables in a small contingency table, usually a 2x2 table. In Honors Statistics, you use it when expected counts are too small for a chi-square test.
Fisher's Exact Test is a significance test for a contingency table in Honors Statistics, usually when you are comparing two categorical variables and the sample is too small for the chi-square approximation to feel safe. The usual setup is a 2x2 table, such as treatment vs. placebo and improved vs. not improved, where one or more expected counts are tiny.
Instead of relying on a large-sample approximation, Fisher's Exact Test calculates the exact probability of getting the observed table, or one that is even more extreme, assuming the null hypothesis of independence is true. That is why it is called exact. The test uses the hypergeometric distribution, which fits situations where you are counting outcomes from a fixed population without replacement.
A good way to picture it is that Fisher's test asks, "If the row and column totals stay fixed and the variables really are independent, how surprising is this particular arrangement of counts?" The answer becomes a p-value. If the p-value is small, the table is unlikely under independence, so you have evidence of an association between the variables.
This is different from just looking at the raw counts. A table with 3 versus 1 in one cell can look dramatic, but the real question is whether that pattern could happen often just by chance given the sample size and the marginal totals. Fisher's Exact Test handles that by working with the exact distribution of possible tables, not a rough normal-based shortcut.
In class, you usually meet Fisher's Exact Test right after or alongside the chi-square test of independence. The main decision point is simple: if the expected cell frequencies are too small, especially in a 2x2 table, Fisher's Exact Test is the safer choice. For bigger tables, or when expected counts are reasonably large, chi-square is usually the method you see first.
Fisher's Exact Test matters because it gives you a trustworthy way to test association when your data are sparse. In Honors Statistics, that comes up any time a sample is small, a category is rare, or a contingency table has counts that make the chi-square approximation shaky.
It also teaches a bigger lesson about statistical methods: the test should match the data, not the other way around. A result from a tiny table can be misleading if you force a chi-square test onto it just because it is the one you remember first. Fisher's test shows that exact probability methods can be better when the usual large-sample shortcuts break down.
This term also connects to how you read outputs and justify conclusions. You are not just saying "there is a relationship" because the counts look different. You are checking whether the pattern is surprising under independence, then explaining what that means in context. That kind of interpretation shows up in homework, labs, and written responses where you have to defend why one test fits the situation better than another.
Keep studying Honors Statistics Unit 4
Visual cheatsheet
view galleryHypergeometric Distribution
Fisher's Exact Test is built on the hypergeometric distribution. That distribution models sampling without replacement, which matches the way the test treats the fixed row and column totals in a contingency table. If you understand why a hypergeometric setup works, Fisher's p-value makes a lot more sense because it is counting exact table probabilities, not approximating them.
Chi-Square Test of Independence
These two tests answer a similar question about whether categorical variables are associated, but they are used in different conditions. Chi-square works well when expected counts are large enough for the approximation to be reasonable. Fisher's Exact Test is the backup when those expected counts are too small, especially in a 2x2 table.
Contingency Table
You usually see Fisher's Exact Test in a contingency table, which organizes counts for two categorical variables. The table structure matters because the test uses the row totals and column totals to evaluate how unusual the observed arrangement is. If you cannot build the table cleanly, you usually cannot run the test correctly.
Nonparametric Test
Fisher's Exact Test is nonparametric, so it does not depend on a normal distribution or a population parameter model in the usual way. That makes it a strong choice for categorical data with small sample sizes. In Stats, this is a useful reminder that not every inference method depends on the same assumptions.
A quiz or lab question may give you a small 2x2 table and ask which test to use, or ask you to explain why chi-square is not a good choice. Your job is to spot the small expected counts and choose Fisher's Exact Test instead of forcing the chi-square test of independence. If the question gives a p-value, you should interpret it in context, then decide whether the data give evidence of an association between the two categorical variables. In a written response, it helps to say that the test is exact because it uses the hypergeometric distribution and does not rely on the large-sample chi-square approximation.
These are the two most common tests for association in categorical data, so they get mixed up a lot. Use chi-square when expected cell counts are large enough, and use Fisher's Exact Test when the table is small or expected counts are too low. Fisher's is exact for small tables, while chi-square is an approximation that works better with larger samples.
Fisher's Exact Test checks whether two categorical variables are independent, usually in a small 2x2 contingency table.
It is called exact because it uses the exact probability of the observed table, or a more extreme one, under the null hypothesis.
The test is based on the hypergeometric distribution, which fits sampling without replacement from a fixed population.
You usually choose Fisher's Exact Test when expected counts are too small for the chi-square test of independence to be reliable.
The conclusion is about association, not causation, so you still have to interpret the result in the context of the data collection.
It is a hypothesis test for checking whether two categorical variables are independent, usually when the sample size is small. You often see it with 2x2 contingency tables. Instead of using a chi-square approximation, it calculates an exact p-value from the table's possible arrangements.
Use Fisher's Exact Test when expected cell counts are too small for the chi-square test of independence to be reliable. This is most common in small 2x2 tables. If the counts are larger, chi-square is usually the method your class expects first.
It is called exact because it computes the precise probability of the observed table, or one more extreme, under the null hypothesis of independence. It does not depend on a large-sample approximation. That is what makes it useful when the data are sparse.
It is most common in 2x2 tables, which is the version many stats classes focus on. It can be extended to larger tables, but that is less common in an introductory or honors course. For most class problems, think small categorical tables with low expected counts.