01 / 21
\(p\)

Setting Up Tests for
Population Proportion

Learn how to formulate null and alternative hypotheses for a population proportion, choose the correct test direction, and verify conditions before running a one-proportion \(z\)-test.

AP Statistics

What Is a Hypothesis Test?

A hypothesis test is a formal procedure for deciding whether sample data provide convincing evidence against a claim about a population parameter. We start with an assumption (the null hypothesis) and ask: "Is the data we observed unlikely enough to reject that assumption?"

Key Vocabulary

\(H_0\) : Null Hypothesis
The default claim we assume is true. Always states \(p = p_0\).
\(H_a\) : Alternative Hypothesis
What we suspect is true instead. Can be \(<\), \(>\), or \(\neq\) \(p_0\).
\(\hat{p}\) : Sample Proportion
The proportion observed in our sample: \(\hat{p} = \frac{X}{n}\).

The Big Picture: Test Flow

State \(H_0\) and \(H_a\) Check Conditions Calculate \(z\)-statistic Conclude \(p\)-value THIS LESSON

The Null Hypothesis \(H_0\)

The null hypothesis is a statement of "no change" or "no difference" from a claimed value. For a one-proportion \(z\)-test, it always takes the form:

The null hypothesis uses equality only. It never uses \(<\), \(>\), or \(\neq\). Think of it as the "status quo" that we try to find evidence against.

The Alternative Hypothesis \(H_a\)

The alternative hypothesis states what we suspect or hope to show. It determines the direction of the test:

Left-Tailed
\(H_a : p < p_0\)
Evidence that \(p\) is less than claimed.
Right-Tailed
\(H_a : p > p_0\)
Evidence that \(p\) is greater than claimed.
Two-Tailed
\(H_a : p \neq p_0\)
Evidence that \(p\) is different from claimed.

One-Tailed vs Two-Tailed

Left-Tailed \(p\)-value \(p_0\) Two-Tailed \(\frac{p\text{-value}}{2}\) \(\frac{p\text{-value}}{2}\) \(p_0\)

Conditions for Inference

Before performing a one-proportion \(z\)-test, you must verify three conditions:

1
Random: The data come from a random sample or randomized experiment.
2
10% Condition (Independence): The sample size \(n\) is no more than 10% of the population: \(n \leq 0.10N\).
3
Large Counts: Both \(np_0 \geq 10\) and \(n(1 - p_0) \geq 10\) so the sampling distribution is approximately Normal.

The \(z\)-Test Statistic

Once the hypotheses are stated and conditions are checked, the test statistic measures how far \(\hat{p}\) is from \(p_0\) in standard-error units:

CI vs Hypothesis Test

Uses \(p_0\): \(\sqrt{\frac{p_0(1-p_0)}{n}}\)
Confidence Interval Hypothesis Test
Purpose Estimate a range for \(p\) Test a claim about \(p\)
Standard Error Uses \(\hat{p}\): \(\sqrt{\frac{\hat{p}(1-\hat{p})}{n}}\)
Large Counts \(n\hat{p} \geq 10\), \(n(1-\hat{p}) \geq 10\) \(np_0 \geq 10\), \(n(1-p_0) \geq 10\)
Result An interval: \((\hat{p} \pm z^* \cdot SE)\) A \(p\)-value

How to Set Up the Test

1
Define the parameter: Let \(p\) = the true proportion of [population] that [has characteristic].
2
State the hypotheses: \(H_0 : p = p_0\) and choose \(H_a\) based on the research question (\(<\), \(>\), or \(\neq\)).
3
Choose significance level: Typically \(\alpha = 0.05\) unless otherwise stated.
4
Check conditions: Random, 10% condition, Large Counts (\(np_0 \geq 10\), \(n(1 - p_0) \geq 10\)).
5
Identify the test: State: "We will perform a one-proportion \(z\)-test."

Example 1: Coin Fairness

A student flips a coin 200 times and gets 115 heads. She wants to test whether the coin is unfair (not 50% heads).

1
Parameter: Let \(p\) = the true proportion of heads for this coin.
2
Hypotheses: \(H_0 : p = 0.50\) vs. \(H_a : p \neq 0.50\) (two-tailed, since "unfair" means different).
3
Conditions: Random flips ✓ | \(np_0 = 200(0.50) = 100 \geq 10\) ✓ | \(n(1-p_0) = 100 \geq 10\) ✓

Example 2: Drug Effectiveness

A pharmaceutical company claims that at least 80% of patients recover with their new drug. A researcher surveys 150 patients and finds 108 recovered. She suspects the true rate is lower than claimed.

1
Parameter: Let \(p\) = the true proportion of patients who recover with the drug.
2
Hypotheses: \(H_0 : p = 0.80\) vs. \(H_a : p < 0.80\) (left-tailed, because "lower").
3
Large Counts: \(np_0 = 150(0.80) = 120 \geq 10\) ✓ | \(n(1-p_0) = 150(0.20) = 30 \geq 10\) ✓

Example 3: Voter Support

A political analyst claims that more than 60% of voters support a new policy. A random poll of 400 voters finds 256 in support.

1
Parameter: Let \(p\) = the true proportion of all voters who support the policy.
2
Hypotheses: \(H_0 : p = 0.60\) vs. \(H_a : p > 0.60\) (right-tailed, since "more than").
3
Large Counts: \(np_0 = 400(0.60) = 240 \geq 10\) ✓ | \(n(1 - p_0) = 400(0.40) = 160 \geq 10\) ✓

Common Mistakes

✕ Using \(\hat{p}\) in \(H_0\)
The null hypothesis is about the population parameter \(p\), not the sample statistic \(\hat{p}\). Always write \(H_0 : p = p_0\).
✕ Wrong Inequality in \(H_a\)
"Greater than claimed" means \(H_a : p > p_0\), not \(H_a : p < p_0\). Read the research question carefully for direction words.
✕ Using \(\hat{p}\) in Large Counts
Check conditions with \(p_0\): verify \(np_0 \geq 10\) and \(n(1 - p_0) \geq 10\). Using \(\hat{p}\) applies to confidence intervals, not hypothesis tests.

Summary of Test Setup

Component What to Write Key Reminder
Parameter \(p\) = true proportion of ... Use context
\(H_0\) \(H_0 : p = p_0\) Always uses \(=\)
\(H_a\) \(<\), \(>\), or \(\neq\) Match the question
Significance \(\alpha = 0.05\) (default) State before data
Conditions Random, 10%, Large Counts Use \(p_0\) not \(\hat{p}\)
Test Name One-proportion \(z\)-test Always state it

Question 1 of 5

A company claims that 90% of its packages arrive on time. A consumer group suspects the true rate is lower. Which hypotheses are correct?

Question 2 of 5

For a one-proportion \(z\)-test with \(H_0 : p = 0.40\) and \(n = 80\), which Large Counts check is correct?

Question 3 of 5

A school principal believes that more than 75% of students eat breakfast. In a random sample of 120 students, 96 eat breakfast. What is the sample proportion \(\hat{p}\)?

Question 4 of 5

A researcher tests \(H_0 : p = 0.50\) vs. \(H_a : p > 0.50\) with \(n = 60\) and \(\hat{p} = 0.60\). What is the value of the test statistic \(z\)?

Question 5 of 5

A survey of 50 people from a town of 300 tests whether the proportion who recycle is different from 0.60. Which condition is not satisfied?