z-Test for Proportions
What you'll learn: How to test hypotheses about a population proportion using the normal approximation to the binomial distribution.
The Core Idea
You've learned to test hypotheses about population means; now we do the same for proportions. When you want to know if a population proportion (like the percentage of voters favoring a candidate, or the defect rate in manufacturing) equals a specific value, you use a z-test for proportions.
The key insight: when the sample size is large enough, the sampling distribution of a sample proportion (where is the count of successes) is approximately normal by the Central Limit Theorem. This lets us use the standard normal distribution to conduct our test.
Setting Up the Test
Hypotheses:
- Null hypothesis: (the population proportion equals some claimed value)
- Alternative: (or or )
Test Statistic:
Notice we use (the hypothesized value) in the standard error, not —this is because under , we assume is true.
When it works: You need and for the normal approximation to be valid.
Carrying Out the Test
- Calculate your sample proportion
- Compute the z-statistic using the formula above
- Find the p-value from the standard normal distribution
- Compare to your significance level and make a decision
Example: If a company claims 90% customer satisfaction and you survey 200 customers finding 170 satisfied (), you can test whether the true proportion really is 0.90.
Key Takeaway: The z-test for proportions uses the normal approximation to test whether a population proportion equals a hypothesized value, standardizing the difference between sample and hypothesized proportions by the appropriate standard error.