Justifying a Claim Based on a Confidence Interval for a Population Proportion - AP Statistics

SECTION 1

Historical Context & Motivation

Long before modern polling, governments and scientists grappled with a fundamental challenge: how to make reliable claims about an entire population when only a fraction of it can be observed. The desire to quantify uncertainty—rather than simply guess—drove mathematicians and statisticians to develop formal frameworks for statistical inference. The confidence interval emerged as one of the most powerful and widely adopted tools in this arsenal, providing a range of plausible values for an unknown parameter based on observed data. Understanding how to use a confidence interval to justify—or refute—a claim about a population proportion is central to evidence-based reasoning in statistics, science, public policy, and business.

1713 • Bernoulli's Ars Conjectandi

Jacob Bernoulli published a foundational theorem showing that sample proportions converge to the true population proportion as sample size grows, establishing the theoretical bedrock for proportion-based inference.

1812 • Laplace's Normal Approximation

Pierre-Simon Laplace formalized the use of the normal distribution to approximate the sampling distribution of proportions, enabling practical interval calculations long before modern computing.

1934 • Neyman's Confidence Interval Framework

Jerzy Neyman introduced the formal concept of a confidence interval, defining the repeated-sampling interpretation that the AP Statistics curriculum uses today. His framework shifted the focus from single estimates to intervals with a guaranteed long-run capture rate.

1960s–Present • Modern Polling and Evidence-Based Policy

Confidence intervals for proportions became the standard reporting tool for election polls, clinical drug trials, and quality-control audits, making the skill of interpreting intervals essential for informed citizens and decision-makers.

The central question this lesson addresses is deceptively straightforward: given a confidence interval for a population proportion p, how do we determine whether a specific claimed value of p is supported or contradicted by the data? Answering this question requires understanding both the construction and the interpretation of the interval, and it forms a bridge between confidence intervals and hypothesis testing—two pillars of statistical inference.

SECTION 2

Core Principles & Definitions

Before using a confidence interval to evaluate a claim, you need a precise understanding of what the interval represents and the logic that connects it to decision-making. The following principles form the conceptual foundation for justifying claims about a population proportion.

Population Proportion (p)

The true fraction of individuals in the entire population who possess a characteristic of interest. This parameter is typically unknown and is the target of inference.

Sample Proportion (p̂)

The observed fraction of successes in a random sample, calculated as p̂ = X / n, where X is the number of successes and n is the sample size. It serves as the point estimate for p.

Confidence Interval

An interval of the form (p̂ − ME, p̂ + ME) constructed so that, across many repeated samples, a specified percentage (the confidence level) of such intervals would contain the true p.

Confidence Level (C)

The long-run proportion of intervals (e.g., 95%) that capture the true parameter when the procedure is repeated. It is NOT the probability that p lies in one specific computed interval.

Justification Logic

If a claimed value of p falls inside the interval, the data are consistent with the claim. If it falls outside, we have convincing evidence at the corresponding significance level that the claim is not plausible.

✦

KEY TAKEAWAY

Think of a confidence interval as a net cast into the ocean to capture a specific fish—the true population proportion. If a claimed value of p lands inside the net, the data 'caught' a range of values that includes the claim, so there is no evidence against it at that confidence level. If the claimed value lands outside the net, the data provide evidence that the true proportion differs from the claim. The width of the net is determined by the confidence level and the sample size—a wider net (higher confidence or larger sample variability) is more likely to include the fish, but less precise.

SECTION 3

Visual Explanation

The diagram below illustrates the core reasoning process. A confidence interval is displayed as a horizontal segment on the number line. Two claimed values of the population proportion are shown: one that falls inside the interval (consistent with the data) and one that falls outside (contradicted by the data). The position of a claimed value relative to the interval boundary is the entire basis for justifying or rejecting the claim.

The cyan-shaded region represents the 95% confidence interval (0.43, 0.61) centered on the sample proportion p̂ = 0.52. The green dashed line shows a claimed value of p₀ = 0.50 that falls inside the interval, meaning the data are consistent with that claim. The red dashed line shows a claimed value of p₀ = 0.68 that falls outside the interval, indicating convincing evidence against that claim at the 5% significance level.

Notice that the justification is binary with respect to any given claimed value: either the value is inside the interval or it is not. However, the strength of the evidence depends on how far outside the interval the claimed value lies and on the chosen confidence level. A claimed value that barely falls outside a 90% interval carries far less evidential weight than one that falls well outside a 99% interval. On the AP exam, you will most often encounter 95% confidence intervals, which correspond to a two-sided significance level of α = 0.05.

SECTION 4

Mathematical Framework

The one-sample z-interval for a population proportion is derived from the sampling distribution of p̂. Under appropriate conditions, the sampling distribution of p̂ is approximately normal with mean p and standard deviation √(p(1 − p)/n). Because p is unknown, we estimate the standard error using p̂, yielding the following interval formula.

CONFIDENCE INTERVAL FOR p

p̂ ± z* × √( p̂(1 − p̂) / n )

where p̂ is the sample proportion, n is the sample size, and z* is the critical value from the standard normal distribution corresponding to the chosen confidence level (e.g., z* = 1.96 for 95% confidence).

MARGIN OF ERROR (ME)

ME = z* × √( p̂(1 − p̂) / n )

The margin of error quantifies the half-width of the interval. Larger samples decrease ME, while higher confidence levels increase it through a larger z*.

Conditions for Inference

✓ Random: The data come from a random sample or a randomized experiment.
✓ 10% Condition (Independence): When sampling without replacement, the sample size n is no more than 10% of the population size N.
✓ Large Counts: Both np̂ ≥ 10 and n(1 − p̂) ≥ 10, ensuring the normal approximation to the binomial is reasonable.

JUSTIFICATION DECISION RULE

If p₀ ∈ (p̂ − ME, p̂ + ME), the data do not provide convincing evidence against p = p₀ at significance level α = 1 − C.

Conversely, if p₀ ∉ (p̂ − ME, p̂ + ME), we have convincing evidence at level α that the true proportion differs from p₀. This mirrors the outcome of a two-sided hypothesis test of H₀: p = p₀.

ℹ

Connection to Hypothesis Testing

A 95% confidence interval contains exactly those values of p₀ for which a two-sided z-test at α = 0.05 would fail to reject H₀: p = p₀. Values outside the interval correspond to values for which the test would reject H₀. This duality means a confidence interval can simultaneously evaluate infinitely many hypotheses.

SECTION 5

Interpreting Results & Common Scenarios

On the AP Statistics exam, questions about justifying a claim often present a confidence interval and a specific claimed value, then ask whether the data support or refute the claim. The following diagram and table categorize the three most common scenarios you will encounter, along with the precise language expected in your response.

The three scenarios show the claimed proportion p₀ relative to the confidence interval. In Scenario A, p₀ lies within the interval, so the claim is plausible. In Scenarios B and C, p₀ falls outside—either below or above—the interval, providing convincing evidence against the claim. Notice that Scenarios B and C also give directional information about how p differs from p₀.

Scenario	Position of p₀	Conclusion	Directional Claim
A	Inside the interval	No convincing evidence that p ≠ p₀	None — the data are consistent with p₀
B	Below the lower bound	Convincing evidence that p ≠ p₀	Evidence that p > p₀
C	Above the upper bound	Convincing evidence that p ≠ p₀	Evidence that p < p₀

Summary of justification outcomes based on position of claimed value relative to the confidence interval

💡

AP Exam Language Tip

The AP scoring rubrics consistently reward precise phrasing. Avoid saying "we are 95% confident that p equals 0.52." Instead, write: "We are 95% confident that the true proportion of [context] is between [lower bound] and [upper bound]." Then state whether the claimed value falls inside or outside this interval and draw the appropriate conclusion in context.

SECTION 6

Worked Example

A school district claims that 60% of its high school graduates enroll in a four-year college. A researcher suspects this claim may be too high and surveys a simple random sample of 250 recent graduates, finding that 134 enrolled in a four-year college. Construct a 95% confidence interval for the true proportion and use it to evaluate the district's claim.

Evaluating the District's Enrollment Claim

1
Step 1 — Identify the Parameter and Claim

Let p represent the true proportion of all recent high school graduates in this district who enroll in a four-year college. The district claims p = 0.60. We want to determine whether the sample data provide convincing evidence against this claim.
2
Step 2 — Check Conditions

Random: The problem states a simple random sample was used. ✓ 10% Condition: It is reasonable that the district has at least 2,500 recent graduates, so 250 ≤ 10% of N. ✓ Large Counts: np̂ = 250 × 0.536 = 134 ≥ 10 and n(1 − p̂) = 250 × 0.464 = 116 ≥ 10. ✓
3
Step 3 — Compute the Sample Proportion

p̂ = 134 / 250 = 0.536.

p̂ = 0.536
4
Step 4 — Calculate the Margin of Error

For a 95% confidence level, z* = 1.96. The standard error is SE = √(0.536 × 0.464 / 250) = √(0.248704 / 250) = √0.000995 ≈ 0.03154. Therefore, ME = 1.96 × 0.03154 ≈ 0.0618.

ME ≈ 0.062
5
Step 5 — Construct the Interval

The 95% confidence interval is (0.536 − 0.062, 0.536 + 0.062) = (0.474, 0.598).

95% CI: (0.474, 0.598)
6
Step 6 — Justify the Claim

The district's claimed value of p = 0.60 is not contained in the 95% confidence interval (0.474, 0.598) because 0.60 > 0.598. Since the claimed value lies above the upper bound of the interval, we have convincing evidence at the α = 0.05 level that the true proportion of graduates who enroll in a four-year college is less than 0.60. The data do not support the district's claim.

p₀ = 0.60 is outside the CI → Convincing evidence p < 0.60

SECTION 7

Strengths, Limitations & Common Pitfalls

Strengths	Limitations
Evaluates many possible values of p simultaneously—every value inside the interval is plausible, every value outside is not.	Only valid when the conditions (Random, 10%, Large Counts) are met; violations can produce misleading intervals.
Provides a range of plausible values, giving more information than a single reject/fail-to-reject decision.	Does not directly give a p-value; a claim might barely fall outside the interval but still be practically meaningful.
Easy to communicate to non-statisticians; the interval has intuitive appeal as a 'range of likely values.'	Misinterpreted frequently: a 95% CI does NOT mean there is a 95% probability that p lies in this particular interval.
Directional inference is possible when the claimed value falls entirely above or below the interval.	Width depends on sample size; small samples produce wide intervals that may be uninformative.

Strengths and limitations of using confidence intervals to justify claims

⚠️

AVOID THIS MISTAKE

A confidence interval captures the parameter—not the data. Saying "there is a 95% chance that the true proportion is between 0.474 and 0.598" is incorrect because p is a fixed (though unknown) value, not a random variable. The correct interpretation is: "We are 95% confident that the interval (0.474, 0.598) captures the true proportion," which means that 95% of all intervals constructed by this method in repeated sampling would contain p.

SECTION 8

Connection to Hypothesis Testing & Advanced Inference

Using a confidence interval to justify a claim about a population proportion is closely related to performing a formal significance test (z-test for a proportion). The table below makes this connection explicit. Understanding the duality between intervals and tests deepens your statistical reasoning and equips you for more advanced topics such as two-proportion z-intervals, chi-square inference, and Bayesian credible intervals.

Feature	Confidence Interval Approach	Hypothesis Test Approach
Starting point	Compute interval using p̂ and SE based on p̂	State H₀: p = p₀ and compute test statistic z using p₀ in SE
Decision criterion	Is p₀ inside or outside the interval?	Is the p-value ≤ α?
Output	Range of plausible values for p	Single p-value and reject/fail-to-reject decision
Directional info	Yes — can tell if p is likely above or below p₀	Only with a one-sided alternative
Agreement	p₀ outside C% CI → reject at α = 1 − C	p-value ≤ α → reject; same conclusion
Minor difference	SE uses p̂	SE uses p₀ under H₀; conclusions may rarely differ at boundary cases

Duality between confidence intervals and significance tests for a population proportion

Looking ahead, the same logic extends naturally to two-proportion confidence intervals, where you compare two groups by checking whether 0 lies inside the interval for p₁ − p₂. In a chi-square goodness-of-fit test, the reasoning generalizes to multiple categories simultaneously. At the graduate level, Bayesian credible intervals replace the frequentist confidence interval by incorporating prior information, and the interpretation shifts to a direct probability statement about the parameter. However, the core idea—comparing a claimed value to a range of data-supported values—remains the same across all these frameworks.

SECTION 9

Practice Problems

Question 1: CONCEPTUAL

A 90% confidence interval for the proportion of adults in a city who support a new transit policy is (0.52, 0.64). A city official claims that a majority (more than 50%) of adults support the policy. Which statement best justifies a conclusion based on this interval?

A. Because 0.50 is not in the interval, there is convincing evidence that a majority supports the policy.
B. Because the interval contains values above 0.50, we are 90% sure the true proportion is exactly 0.58.
C. Because 0.50 is below the interval, there is convincing evidence that fewer than half of adults support the policy.
D. Because 0.50 is not in the interval, there is a 90% probability that the true proportion is between 0.52 and 0.64.
E. Because the sample proportion is likely around 0.58, the claim that a majority supports the policy cannot be evaluated with a confidence interval.

Show answer (A)

A is correct. The entire interval (0.52, 0.64) lies above 0.50, meaning 0.50 is below the lower bound. This provides convincing evidence at the 10% significance level that the true proportion exceeds 0.50, supporting the majority claim. B is wrong because a confidence interval provides a range of plausible values, not a claim about a single exact value. C is wrong because 0.50 being below the interval means the data support p > 0.50, not p < 0.50—the direction is reversed. D is wrong because saying there is a 90% probability the parameter is in this specific interval is an incorrect interpretation of a frequentist confidence interval. E is wrong because confidence intervals are precisely the tool used to evaluate such claims.

Question 2: BASIC CALCULATION

In a random sample of 400 registered voters, 172 favored a ballot initiative. A political group claims that 40% of all registered voters favor the initiative. The 95% confidence interval for p is (0.382, 0.478). Based on this interval, which conclusion is most appropriate?

A. There is convincing evidence that the true proportion is greater than 0.40 because p̂ = 0.43 > 0.40.
B. There is convincing evidence that the true proportion differs from 0.40 because the interval is wide.
C. There is not convincing evidence that the true proportion differs from 0.40 because 0.40 is contained in the interval.
D. We accept the claim that p = 0.40 because the confidence interval captures this value.
E. The interval provides convincing evidence that exactly 43% of voters favor the initiative.

Show answer (C)

C is correct. Because 0.40 falls within the 95% confidence interval (0.382, 0.478), the data do not provide convincing evidence that the true proportion differs from 0.40 at the α = 0.05 significance level. The claimed value is a plausible value of p. A is wrong because comparing p̂ to p₀ alone (without referencing the interval) ignores sampling variability; the difference could easily be due to chance. B is wrong because the width of the interval is not the criterion; what matters is whether 0.40 lies inside it. D is wrong because we never 'accept' H₀ in frequentist statistics—we only fail to reject it. E is wrong because a confidence interval provides a range of plausible values, not evidence for a single specific value.

Question 3: INTERMEDIATE

A hospital administrator claims that fewer than 15% of patients discharged in the past year were readmitted within 30 days. A quality team takes a simple random sample of 500 discharged patients and constructs a 95% confidence interval for the proportion readmitted, obtaining (0.104, 0.156). Which conclusion is best supported?

A. The data provide convincing evidence that the readmission rate is less than 15% because the sample proportion is less than 0.15.
B. The data do not provide convincing evidence that the readmission rate differs from 15% because 0.15 is contained in the interval.
C. The data provide convincing evidence that the readmission rate is greater than 15% because the upper bound exceeds 0.15.
D. The data provide convincing evidence that the readmission rate is less than 15% because most of the interval is below 0.15.
E. No conclusion can be drawn because the confidence interval is too narrow.

Show answer (B)

B is correct. The value 0.15 is within the interval (0.104, 0.156), so the data are consistent with the claim that the readmission rate could be 15%. At the α = 0.05 level, there is not convincing evidence that the rate differs from 15%. A is wrong because the sample proportion being below 0.15 does not constitute convincing evidence on its own; the interval (which accounts for sampling variability) still includes 0.15. C is wrong because the fact that the upper bound slightly exceeds 0.15 does not mean the rate is greater than 0.15; it means 0.15 is a plausible value. D is wrong because the proportion of the interval below a value does not constitute the standard for convincing evidence; the criterion is strictly whether the value is inside or outside the interval. E is wrong because a narrow interval is more informative, not less; the interval clearly identifies 0.15 as a plausible value.

Question 4: APPLIED

4 pts

A consumer advocacy group wants to test whether more than 25% of online shoppers in a large metropolitan area have experienced credit card fraud. They take a simple random sample of 600 online shoppers and find that 174 report having experienced fraud. (a) Construct a 95% confidence interval for the population proportion of online shoppers who have experienced credit card fraud. (b) Based on the interval, is there convincing evidence that the proportion exceeds 0.25? Justify your answer. (c) If the sample size were increased to 2,400 (with the same sample proportion), how would the interval change, and would the conclusion in part (b) necessarily change? Explain.

Show answer

Rubric (4 points): Point 1 — Conditions and Setup: State the parameter: p = true proportion of online shoppers in the metro area who have experienced credit card fraud. Verify conditions: Random (SRS stated ✓), 10% condition (600 ≤ 10% of all online shoppers in a large metro area ✓), Large Counts (np̂ = 174 ≥ 10 and n(1 − p̂) = 426 ≥ 10 ✓). Point 2 — Interval Construction: p̂ = 174/600 = 0.29. SE = √(0.29 × 0.71 / 600) = √(0.000343) ≈ 0.01852. ME = 1.96 × 0.01852 ≈ 0.0363. The 95% CI is (0.29 − 0.0363, 0.29 + 0.0363) = (0.254, 0.326). Point 3 — Justification of Claim: The value 0.25 is below the lower bound of 0.254, so 0.25 is not contained in the interval. Because all plausible values of p exceed 0.25, there is convincing evidence at the α = 0.05 level that more than 25% of online shoppers have experienced credit card fraud. Point 4 — Effect of Larger Sample: With n = 2,400 and p̂ = 0.29, the SE would be halved (since √(600/2400) = 1/2), yielding SE ≈ 0.00926 and ME ≈ 0.0182. The new interval would be approximately (0.272, 0.308), which is narrower and still entirely above 0.25. The conclusion would not change—in fact, the evidence would be even stronger because the interval would be farther from 0.25. In general, a larger sample makes the interval narrower, so if the claim was already refuted, it will remain refuted; however, this is guaranteed only when the sample proportion stays the same.

Question 5: CRITICAL THINKING

4 pts

A researcher constructs a 95% confidence interval for the proportion of college students who study abroad, obtaining (0.118, 0.202). A university president claims that exactly 20% of students study abroad. (a) Does the confidence interval provide convincing evidence against the president's claim? Explain. (b) Suppose the researcher also constructs a 99% confidence interval using the same sample data. Without performing calculations, explain whether the 99% interval would be more likely, less likely, or equally likely to contain the value 0.20 compared to the 95% interval. Justify your reasoning. (c) A colleague argues: "Since 0.20 is in the interval, we have proven the president's claim is true." Identify and correct the statistical error in this reasoning. (d) Explain how the researcher could use the confidence interval approach and the significance test approach to reach the same conclusion, and identify one situation where the two approaches might lead to different conclusions.

Show answer

Rubric (4 points): Point 1 — Part (a) Justification: The claimed value of p = 0.20 falls inside the 95% confidence interval (0.118, 0.202) because 0.118 < 0.20 < 0.202. Therefore, 0.20 is a plausible value for the true proportion, and the data do not provide convincing evidence against the president's claim at the 0.05 significance level. Point 2 — Part (b) CI Width and Confidence Level: A 99% confidence interval uses a larger critical value (z* ≈ 2.576 compared to 1.96 for 95%), so it will be wider than the 95% interval while sharing the same center at the sample proportion. Since 0.20 is already inside the narrower 95% interval, it must also be inside the wider 99% interval. A higher confidence level produces a wider interval that contains all values from the lower-level interval plus additional values, so the 99% interval is guaranteed to contain 0.20 in this case. Point 3 — Part (c) Error Identification: The colleague commits the error of treating a failure to reject the null hypothesis as proof that the null hypothesis is true. A confidence interval identifies a range of plausible values for the population proportion. The fact that 0.20 is inside the interval means the data are consistent with p = 0.20, but they are also consistent with every other value in the interval, such as p = 0.15 or p = 0.19. We cannot conclude that p is exactly 0.20. We can only say that we lack sufficient evidence to rule it out. In frequentist inference, we never prove a parameter equals a specific value; we only fail to find evidence against it. Point 4 — Part (d) Duality and Discrepancy: Both approaches test whether p = 0.20 is plausible. The confidence interval approach checks whether 0.20 falls inside the interval constructed using the standard error based on the sample proportion, SE = √(p̂(1 − p̂)/n). The hypothesis test approach computes a z test statistic, z = (p̂ − 0.20) / √(0.20 × 0.80 / n), and compares the resulting p-value to α = 0.05. Both generally yield the same conclusion because they are algebraically similar in structure: if 0.20 is inside the confidence interval, the hypothesis test typically fails to reject, and if 0.20 is outside the interval, the test typically rejects. However, the two approaches can occasionally disagree at boundary cases because the confidence interval uses p̂ in the standard error while the hypothesis test uses the null value p₀ = 0.20. When p̂ is far from 0.20, the two standard errors differ enough that a value might fall just inside the confidence interval while the test just barely rejects, or vice versa. This discrepancy is minor and vanishes as the sample size increases.

SUMMARY

Lesson Summary

Summary

A confidence interval for a population proportion is constructed as p̂ ± z* × √(p̂(1 − p̂)/n) after verifying the Random, 10%, and Large Counts conditions. To justify a claim, check whether the claimed value p₀ falls inside or outside this interval. If p₀ is inside the interval, the data do not provide convincing evidence against the claim at the corresponding significance level α = 1 − C. If p₀ is outside the interval, there is convincing evidence that the true proportion differs from the claimed value, with directional information available from whether p₀ lies above or below the interval.

Remember three critical points for the AP exam: (1) a confidence interval gives a range of plausible values for the parameter, not a probability that the parameter is in the interval; (2) we never 'accept' a null hypothesis—we only fail to reject it; and (3) the duality between confidence intervals and hypothesis tests means that a C% interval rejects the same values that a two-sided test at α = 1 − C would reject, with minor exceptions due to different standard error formulas.