Imagine random samples of men and women are asked if they donated blood in the past 12 months. Which test would compare the proportion who donated between genders?

Imagine random samples of men and women are asked if they donated blood in the past 12 months. Which test would compare the proportion who donated between genders?

• A. 2-proportion z-test
• B. Chi-square test of independence
• C. Fisher's exact test
• D. Paired t-test

Answer: (C)

When you compare the proportion of people who donated between two independent groups (men and women) on a binary outcome (donated yes/no), you’re testing whether the two proportions differ. Fisher’s exact test is the precise choice for this 2x2 situation because it computes the exact probability of the observed distribution under the assumption of independence, without relying on large-sample approximations. This makes it especially reliable when the sample sizes are small or any expected cell count is low. If you had large samples, a two-proportion z-test would be common, since it relies on a normal approximation. But with smaller samples, the normal approximation can be inaccurate, which is why Fisher’s exact test is preferred in this context. The paired t-test isn’t appropriate here because it’s for comparing means in paired measurements, not proportions. The chi-square test of independence is similar in purpose but relies on an approximation that’s less reliable with small counts, whereas Fisher’s exact test gives an exact result.

When you compare the proportion of people who donated between two independent groups (men and women) on a binary outcome (donated yes/no), you’re testing whether the two proportions differ.

Fisher’s exact test is the precise choice for this 2x2 situation because it computes the exact probability of the observed distribution under the assumption of independence, without relying on large-sample approximations. This makes it especially reliable when the sample sizes are small or any expected cell count is low.

If you had large samples, a two-proportion z-test would be common, since it relies on a normal approximation. But with smaller samples, the normal approximation can be inaccurate, which is why Fisher’s exact test is preferred in this context. The paired t-test isn’t appropriate here because it’s for comparing means in paired measurements, not proportions. The chi-square test of independence is similar in purpose but relies on an approximation that’s less reliable with small counts, whereas Fisher’s exact test gives an exact result.