Constructing a Confidence Interval for a Proportion — AP Statistics Study Guide

For: AP Statistics candidates sitting AP Statistics.

Covers: One-proportion z-interval formula, conditions for inference for proportions, interpreting confidence intervals, common critical z-values, margin of error calculation, sample size planning for a desired margin of error, and step-by-step interval construction for a single population proportion.

You should already know: Sampling distributions for sample proportions, basic properties of the normal distribution, point estimation for unknown population parameters.

A note on the practice questions: All worked questions in the "Practice Questions" section below are original problems written by us in the AP Statistics style for educational use. They are not reproductions of past College Board / Cambridge / IB papers and may differ in wording, numerical values, or context. Use them to practise the technique; cross-check with official mark schemes for grading conventions.

1. What Is Constructing a Confidence Interval for a Proportion?

This topic is part of Unit 6: Inference for Categorical Data: Proportions, which makes up 12-15% of the AP Statistics exam weight per the official Course and Exam Description (CED). It appears in both multiple choice (MCQ) and free response (FRQ) sections, most often as a standalone MCQ or the first two parts of a multi-part FRQ. A confidence interval for a proportion gives a range of plausible values for the unknown fixed population proportion (p), which is the proportion of individuals or items in a population that have a specific categorical characteristic.

Standard notation conventions: (\hat{p}) (p-hat) denotes the sample proportion, our point estimate of (p); (n) is sample size; (z^*) is the critical z-value matching our chosen confidence level. It is also commonly called a one-proportion z-interval. Unlike a single point estimate, a confidence interval explicitly quantifies the uncertainty from random sampling, telling us how much our estimate might vary from the true population value. On the AP exam, you will be expected to check conditions, construct the interval, interpret it in context, or calculate required sample size for a desired margin of error.

2. Conditions for Inference for a One-Proportion Confidence Interval

Before constructing any confidence interval, you must verify three core conditions to ensure your sampling distribution of (\hat{p}) is approximately normal, your sample is representative, and your inference is statistically valid. Skipping conditions is the most common cause of lost points on AP FRQs, so this step is non-negotiable.

The three required conditions are:

1. Random: Data must come from a random sample of the population of interest, or a randomized experiment. This ensures your sample is not systematically biased.
2. Independent: Individual observations must be independent. When sampling without replacement from a finite population, this is verified via the 10% condition: the sample size (n) must be no more than 10% of the total population size (N) ((n \leq 0.1N)).
3. Normal/Large Sample: The sampling distribution of (\hat{p}) is approximately normal, which requires at least 10 successes and 10 failures in your sample: (n\hat{p} \geq 10) and (n(1-\hat{p}) \geq 10). (The AP CED uses 10 as the threshold, so stick with this for the exam.)

Worked Example

Problem: A local animal shelter wants to estimate the proportion of adopted dogs that are returned within the first 30 days. They randomly pull records for 62 adoptions from the last 5 years, which has a total of 780 adoptions. 8 adoptions resulted in a return within 30 days. Check all conditions for constructing a confidence interval for the true proportion of returns.

Solution:

1. Random: The problem explicitly states the sample of adoption records was randomly selected, so the Random condition is satisfied.
2. Independent: Total population of adoptions is 780. 10% of 780 is 78, and the sample size 62 ≤ 78, so the 10% condition is met, and the Independence condition is satisfied.
3. Normal: Number of successes (returns) is 8, so (n\hat{p} = 8), which is less than 10. The Normal condition is not satisfied.

In this case, we cannot proceed with a one-proportion z-interval because the large sample condition fails.

Exam tip: On AP FRQs, you must explicitly name each condition and verify it with context-specific numbers. Never just write "conditions are met"—you will lose at least 1 point for missing the numerical checks.

3. Step-by-Step Construction of a One-Proportion Z-Interval

Once conditions are verified, a confidence interval follows the general structure: point estimate ± margin of error. For a population proportion, the point estimate is the sample proportion (\hat{p} = \frac{\text{number of successes}}{n}). The margin of error (ME) is the product of the critical value (z*) (which depends only on your chosen confidence level) and the standard error of (\hat{p}). The standard error (SE) estimates the standard deviation of the sampling distribution of (\hat{p}), since the true population proportion (p) is unknown.

The formula for the one-proportion z-confidence interval is: $$\hat{p}\pm z^{\ast }\sqrt{\frac{\hat{p}(1-\hat{p})}{n}}$$ Common values for (z^) that you should memorize are: 1.645 for 90% confidence, 1.96 for 95% confidence, and 2.576 for 99% confidence. The intuition behind the formula is that the interval is centered at our best guess (the sample proportion), and the margin of error accounts for sampling variability. Higher confidence requires a larger (z^), leading to a wider interval, which makes sense: we need a larger range to be more confident we captured the true proportion.

Worked Example

Problem: A coffee shop wants to estimate the proportion of customers who order oat milk instead of dairy. They randomly sample 120 customers, and 42 order oat milk. Construct a 95% confidence interval for the true proportion of all customers who order oat milk, after verifying conditions are met.

Solution:

1. Conditions are already verified to be met.
2. Calculate the point estimate: (\hat{p} = \frac{42}{120} = 0.35).
3. For 95% confidence, the critical value is (z^* = 1.96).
4. Calculate the margin of error: $$ME=1.96\times \sqrt{\frac{(0.35)(0.65)}{120}}=1.96\times \sqrt{0.001896}\approx 0.085$$
5. Construct the interval: (0.35 \pm 0.085 = (0.265, 0.435)).

The 95% confidence interval for the true proportion is (0.27, 0.44).

Exam tip: Do not round intermediate values like (\hat{p}) or (SE) to fewer than 3-4 decimal places. Rounding early can change your final margin of error enough to lose points for an incorrect interval.

4. Interpreting Confidence Intervals and Confidence Levels

Interpretation questions are extremely common on the AP exam, and require specific, context-rich wording to earn full credit. Many students misinterpret what "C% confidence" actually means, so it is important to memorize the correct phrasing.

• Interpretation of a confidence interval: "We are [C]% confident that the true proportion of [context of your population parameter] lies between [lower bound] and [upper bound]."
• Interpretation of a confidence level: The confidence level describes the long-run behavior of the method: if we took many, many random samples of the same size from the same population, and constructed a confidence interval for each sample, approximately C% of those intervals would contain the true population proportion.

A common misconception is that "95% confidence" means there is a 95% probability the true proportion is in the interval. This is incorrect: the true proportion is a fixed, unknown value, so it is either in the interval or not. The probability describes the method of interval construction, not the position of the fixed true proportion.

Worked Example

Problem: Interpret the 95% confidence interval (0.265, 0.435) from the previous coffee shop example, and interpret what "95% confidence" means in this context.

Solution:

1. Interval interpretation: We are 95% confident that the true proportion of all customers at this coffee shop who order oat milk instead of dairy is between 0.27 and 0.44.
2. Confidence level interpretation: If we repeatedly take random samples of 120 customers from this coffee shop and construct a 95% confidence interval for each sample, about 95% of those intervals will capture the true proportion of customers who order oat milk.

Exam tip: Never use the word "probability" when interpreting a confidence interval for a fixed population parameter. AP graders will mark this wrong even if your context is otherwise correct.

5. Calculating Required Sample Size for a Desired Margin of Error

Researchers often want to plan a study to achieve a specific maximum margin of error for a given confidence level. We can rearrange the margin of error formula to solve for the minimum required sample size (n): $$n=\frac{(z^{\ast }{)}^2\hat{p}(1-\hat{p})}{(ME{)}^2}$$ If you have a prior estimate of (\hat{p}) from a previous study, use that value. If you do not have a prior estimate, use (\hat{p} = 0.5) to get the most conservative (largest) sample size. This works because the product (\hat{p}(1-\hat{p})) is maximized when (\hat{p} = 0.5), so using 0.5 guarantees that your resulting sample size will produce a margin of error no larger than your desired value. A key rule for sample size calculation: always round up to the next whole number, even if the decimal part is less than 0.5, because a fraction of an individual is impossible, and rounding down would give a margin of error slightly larger than desired.

Worked Example

Problem: The coffee shop owner wants to redo their study to get a margin of error of no more than 5% (0.05) for 95% confidence. They use the prior estimate of 0.35 from the first study. What is the minimum sample size they need? What would it be if they had no prior estimate?

Solution:

1. For 95% confidence, (z^* = 1.96), (ME = 0.05), (\hat{p} = 0.35).
2. Plug into the formula: $$n=\frac{(1.96{)}^2(0.35)(0.65)}{(0.05{)}^2}=\frac{3.8416\times 0.2275}{0.0025}\approx 349.6$$
3. Round up to get (n = 350) with the prior estimate.
4. With no prior estimate, use (\hat{p} = 0.5): $$n=\frac{(1.96{)}^2(0.5)(0.5)}{(0.05{)}^2}=384.16$$ Round up to get (n = 385).

Exam tip: If the question does not mention a prior estimate, you must use (\hat{p} = 0.5) to get full credit. Do not guess a value for (\hat{p}) if one is not given.

6. Common Pitfalls (and how to avoid them)

• Wrong move: Stating "there is a 95% probability the true proportion is in our interval" when interpreting a confidence interval. Why: Students confuse the variability of the sampling method with the fixed nature of the true population parameter; the true (p) is not random. Correct move: Interpret the confidence level as describing the long-run proportion of intervals that capture the true parameter, and interpret the interval as a range of plausible values for the fixed (p) with the correct wording.
• Wrong move: Skipping numerical checks for conditions, just writing "conditions are met" on an FRQ. Why: Students think it is just a formality, but AP graders require explicit verification to earn points. Correct move: For each condition, state the condition name, plug in the relevant numbers from the problem, and confirm it is satisfied.
• Wrong move: Using the population proportion (p) instead of the sample proportion (\hat{p}) when calculating the standard error for a confidence interval. Why: Students confuse the standard deviation of the sampling distribution (which uses (p)) with the standard error (which estimates it with (\hat{p}), since (p) is unknown). Correct move: Always use (\hat{p}) in the standard error formula when constructing a confidence interval for an unknown (p).
• Wrong move: Rounding down the required sample size when the result is a non-integer. Why: Students follow standard rounding rules, which is incorrect for sample size planning. Correct move: Always round up to the next whole number, regardless of the decimal part, to ensure the margin of error is no larger than desired.
• Wrong move: Not using (\hat{p} = 0.5) when no prior estimate of (p) is given for sample size calculation. Why: Students forget that 0.5 maximizes the product (\hat{p}(1-\hat{p})) to get the conservative sample size. Correct move: If no prior estimate of (p) is provided, always use (\hat{p} = 0.5) to calculate the required sample size.
• Wrong move: Claiming that a confidence interval of (0.27, 0.44) means 95% of sample proportions lie in this interval. Why: Students confuse the purpose of the interval, which estimates the true population proportion, not the distribution of sample proportions. Correct move: Always frame the interval around the true unknown population proportion, not sample statistics.

7. Practice Questions (AP Statistics Style)

Question 1 (Multiple Choice)

A researcher estimates the proportion of U.S. adults who have hiked in the last year, and constructs a 90% confidence interval of (0.48, 0.56). Which of the following statements is correct? A) 90% of U.S. adults have hiked between 48% and 56% of the time in the last year. B) There is a 90% probability that the true proportion of U.S. adults who hiked in the last year is between 0.48 and 0.56. C) We are 90% confident that the true proportion of U.S. adults who hiked in the last year is between 0.48 and 0.56. D) If we take 100 samples, 90 of them will have a sample proportion between 0.48 and 0.56.

Worked Solution: We need to identify the correct interpretation of a confidence interval. Option A misinterprets the parameter; we are estimating the proportion of adults who hiked, not the frequency of hiking for individual adults. Option B uses "probability" to describe the fixed true proportion, which is incorrect. Option D confuses the location of the true parameter with the distribution of sample statistics. Option C matches the correct interpretation of a confidence interval. The correct answer is C.

Question 2 (Free Response)

A high school student council wants to estimate the proportion of students who support extending the lunch period by 10 minutes. They randomly select 100 students from the school's 1600 total students, and 68 support the extension. (a) Check all conditions for constructing a 90% confidence interval for the true proportion. (b) Construct and interpret a 90% confidence interval for the true proportion of all students who support the extension. (c) The principal claims that a majority (more than 50%) of students support the extension. Is this claim plausible based on your interval? Explain.

Worked Solution: (a) 1. Random: The sample is stated to be random, so the Random condition is satisfied. 2. Independent: 10% of the total population (1600) is 160, and 100 ≤ 160, so the 10% condition is met, and Independence is satisfied. 3. Normal: (n\hat{p} = 68 \geq 10), (n(1-\hat{p}) = 32 \geq 10), so the Normal condition is satisfied. All conditions are met. (b) Calculate (\hat{p} = 68/100 = 0.68). For 90% confidence, (z^* = 1.645). Margin of error: $$ME=1.645\times \sqrt{\frac{0.68(0.32)}{100}}\approx 0.077$$ Interval: (0.68 \pm 0.077 = (0.603, 0.757)). Interpretation: We are 90% confident that the true proportion of all students at this school who support extending lunch is between 0.60 and 0.76. (c) The entire interval is above 0.50, so all plausible values for the true proportion are greater than 0.50. Therefore, the principal's claim that a majority of students support the extension is plausible, and in fact all values in the interval support the claim.

Question 3 (Application / Real-World Style)

A public health researcher wants to estimate the proportion of teenagers in a large state that have received the HPV vaccine, with a margin of error of 3% (0.03) and 99% confidence. A prior survey found that 58% of teenagers in the state had received the vaccine. What is the minimum number of teenagers the researcher needs to sample to achieve the desired margin of error?

Worked Solution: For 99% confidence, (z^* = 2.576), desired (ME = 0.03), prior estimate (\hat{p} = 0.58). Plug into the sample size formula: $$n=\frac{(2.576{)}^2(0.58)(0.42)}{(0.03{)}^2}=\frac{6.636\times 0.2436}{0.0009}\approx 1795.6$$ We round up to the next whole number, so (n = 1796). Interpretation: The researcher needs to sample a minimum of 1796 teenagers to get a margin of error no larger than 3% with 99% confidence, using the prior estimate of 58% vaccination rate.

8. Quick Reference Cheatsheet

9. What's Next

Constructing a confidence interval for a single proportion is the foundational inference skill for all inference on categorical proportions, a core tested topic on the AP exam. Next, you will move to hypothesis testing for a single population proportion, which relies on the same conditions for inference and understanding of the sampling distribution of (\hat{p}) you mastered here. Without correctly checking conditions and calculating intervals for a single proportion, you will struggle with the next core topic: comparing two proportions with confidence intervals and hypothesis tests, which is a frequently tested multi-part FRQ topic. This topic also builds on your earlier knowledge of sampling distributions and point estimation, and feeds into the larger course theme of statistical inference, where we use sample data to quantify uncertainty when drawing conclusions about unknown population parameters.

Hypothesis Testing for a Proportion Confidence Intervals for the Difference of Two Proportions Inference for Chi-Square Distributions