Justifying a Claim About a Population Mean Based on a Confidence Interval - AP Statistics

SECTION 1

Historical Context & Motivation

The problem of drawing conclusions about a population from limited sample data has occupied statisticians for over a century. Before the development of formal inferential tools, researchers relied on informal judgment and large-sample approximations that often proved unreliable for the small datasets typical of agricultural, medical, and industrial experiments. The desire to quantify uncertainty in estimation—rather than pretending estimates are exact—drove the creation of the confidence interval, one of the most widely used tools in statistical inference. Understanding this history clarifies why confidence intervals, rather than single point estimates, serve as the primary vehicle for justifying claims about population parameters.

1908 • Student's t-Distribution

William Sealy Gosset, publishing under the pseudonym "Student," derived the t-distribution for small-sample inference at the Guinness Brewery. His work made it possible to construct intervals for population means even with limited data.

1930 • Neyman's Confidence Interval Framework

Jerzy Neyman formalized the concept of a confidence interval, defining it as a procedure whose long-run capture rate of the true parameter equals the stated confidence level. This shifted the focus from any single interval to the reliability of the method itself.

1937 • Neyman's Foundational Paper

Neyman published his landmark paper distinguishing confidence intervals from fiducial intervals. He established the frequentist interpretation that dominates AP Statistics today: in repeated sampling, C% of all constructed intervals will capture the true parameter.

2000s • APA and Modern Reporting Standards

The American Psychological Association and leading journals began requiring confidence intervals alongside or instead of p-values in published research, recognizing that intervals convey both the direction and precision of estimates more transparently than hypothesis tests alone.

The central question this lesson addresses is straightforward yet powerful: given a confidence interval for a population mean, how do we use it to justify or refute a specific claim about that mean? Rather than computing a new test statistic, we can leverage the interval itself—checking whether a claimed value falls inside or outside it—to draw defensible conclusions. This duality between confidence intervals and hypothesis tests is a cornerstone of modern statistical reasoning and appears frequently on the AP Statistics exam.

SECTION 2

Core Principles & Definitions

Before applying confidence intervals to evaluate claims, it is essential to understand the foundational ideas that make such inference valid. These principles govern the construction, interpretation, and logical use of confidence intervals for a population mean μ.

Confidence Interval Definition

A confidence interval is a range of plausible values for the population mean μ, calculated from sample data. A C% confidence interval means that the procedure used to construct the interval captures the true μ in C% of all possible random samples.

Point Estimate ± Margin of Error

Every confidence interval has the form x̄ ± ME, where x̄ is the sample mean and ME is the margin of error. The margin of error quantifies the precision of the estimate and depends on the critical value, the standard deviation, and the sample size.

Claim Evaluation Logic

If a claimed value μ₀ lies outside the confidence interval, the data provide evidence against that claim at the corresponding significance level. If μ₀ lies inside the interval, the data do not provide sufficient evidence to reject the claim.

Duality with Hypothesis Tests

A two-sided hypothesis test at significance level α rejects H₀: μ = μ₀ if and only if μ₀ falls outside the corresponding (1 − α) × 100% confidence interval. This duality allows confidence intervals to serve double duty as both estimation and testing tools.

Conditions for Validity

The interval is valid only when three conditions are met: the sample is collected via a random process, observations are independent (10% condition for sampling without replacement), and the sampling distribution of x̄ is approximately Normal (large n or population is Normal).

✦

KEY TAKEAWAY

Think of a confidence interval as a net cast into the sea of possible parameter values. If a researcher's claimed value of μ falls inside your net, the data are consistent with that claim—much like a fish caught in the net could plausibly be the target species. If the claimed value falls outside the net, your data provide convincing evidence against that claim at the corresponding confidence level. The width of the net corresponds to the margin of error: a narrower net (smaller ME) demands more precision to include a given value, while a wider net includes more possibilities but gives less definitive conclusions.

SECTION 3

Visual Explanation

The diagram below illustrates how a confidence interval is used to evaluate two different claims about a population mean. The interval is centered at the sample mean x̄, and the two claimed values μ₀ are assessed based on whether they fall inside or outside the interval. This visual representation captures the core logic: plausible values are those contained within the interval, and implausible values are excluded by the data at the given confidence level.

The cyan shaded region represents the 95% confidence interval centered at x̄. A claimed value μ₀ = 68 (green) falls inside the interval and is therefore plausible; the data do not provide evidence to reject this claim. A claimed value μ₀ = 60 (red) falls outside the interval, meaning the data provide convincing evidence against this claim at the 5% significance level.

Notice that the decision is binary with respect to a given confidence level: a claimed value either falls inside the interval or it does not. However, values near the boundary are only marginally plausible or implausible. This is why reporting the entire interval is more informative than simply stating a reject/fail-to-reject conclusion. The interval communicates both the estimated value and the range of uncertainty around it, enabling readers to assess claims for themselves.

SECTION 4

Mathematical Framework

The mathematical machinery behind using confidence intervals to justify claims about a population mean involves the construction of the interval itself and the logical connection to two-sided hypothesis testing. We present the key equations for both the z-interval (when σ is known) and the t-interval (the far more common case when σ is unknown), along with the formal decision rule.

ONE-SAMPLE t CONFIDENCE INTERVAL

x̄ ± t* × (s / √n)

where x̄ is the sample mean, t* is the critical value from the t-distribution with df = n − 1, s is the sample standard deviation, and n is the sample size. The quantity s / √n is the standard error of x̄.

MARGIN OF ERROR

ME = t* × (s / √n)

The margin of error determines the half-width of the interval. A larger sample size n decreases ME (more precision), while a higher confidence level increases t* and thus increases ME (wider interval).

DECISION RULE FOR CLAIM EVALUATION

If μ₀ ∈ (x̄ − ME, x̄ + ME), do not reject the claim; if μ₀ ∉ (x̄ − ME, x̄ + ME), reject the claim.

This rule is equivalent to a two-sided hypothesis test H₀: μ = μ₀ vs. Hₐ: μ ≠ μ₀ at significance level α = 1 − C, where C is the confidence level expressed as a proportion.

ONE-SAMPLE z CONFIDENCE INTERVAL (σ KNOWN)

x̄ ± z* × (σ / √n)

Used when the population standard deviation σ is known. In practice this is rare, but it appears in AP Statistics problems where σ is given. Common z* values: 1.645 for 90%, 1.960 for 95%, and 2.576 for 99% confidence.

ℹ

Confidence Level ↔ Significance Level Duality

A C% confidence interval and a two-sided test at α = 1 − C always agree in their conclusions. For example, a 95% CI corresponds to α = 0.05, and a 99% CI corresponds to α = 0.01. If the claimed μ₀ is inside the 95% CI, a two-sided test at α = 0.05 would fail to reject H₀. If μ₀ is outside the 95% CI, the test would reject H₀.

SECTION 5

Interpreting Intervals & Common Misinterpretations

Correct interpretation of a confidence interval is arguably the most heavily tested skill in the inference portion of the AP Statistics exam. Examiners consistently penalize two categories of errors: claiming a probability statement about the specific interval ("there is a 95% chance that μ is in this interval") and confusing the interval's reference to the parameter with a statement about individual observations or sample means. The diagram below contrasts correct and incorrect language.

The left column shows correct interpretation and justification language. The right column shows common errors that will lose points on the AP exam. Note that the 95% refers to the long-run capture rate of the method, not to any single computed interval.

When justifying a claim on the AP exam, a complete response must contain three elements: (1) identify the claimed value μ₀, (2) state whether μ₀ falls inside or outside the given confidence interval, and (3) draw the appropriate conclusion using language tied to the confidence level. Saying "the interval does not contain 60, so we have convincing evidence that the true mean differs from 60" is a model response. Omitting the connection to the interval's content, or making a probability statement about the fixed parameter, will result in deducted rubric points.

Scenario	Claimed μ₀ Position	Conclusion
μ₀ inside the CI	Plausible value	The data do not provide convincing evidence that μ differs from μ₀.
μ₀ outside the CI	Implausible value	The data provide convincing evidence that μ differs from μ₀ (at the corresponding significance level).
μ₀ at the boundary	Borderline	The evidence is marginal; increasing the sample size or choosing a different confidence level may change the conclusion.

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

SECTION 6

Worked Example

A nutritionist claims that the mean daily caloric intake for college students at a large university is 2100 calories. A researcher collects a simple random sample of 36 students and records their daily caloric intake. The sample yields x̄ = 2250 calories and s = 420 calories. Construct a 95% confidence interval for μ and use it to evaluate the nutritionist's claim.

Evaluating a Claim About Mean Caloric Intake

1
Step 1 — State the Parameter and Claim

Let μ represent the true mean daily caloric intake for all college students at this university. The nutritionist's claim is that μ = 2100 calories. We will construct a 95% confidence interval for μ and check whether 2100 falls inside or outside.
2
Step 2 — Check Conditions

Random: The problem states a simple random sample was collected. Independence: The 36 students are less than 10% of all students at a large university, so the 10% condition is satisfied. Normal: n = 36 ≥ 30, so by the Central Limit Theorem, the sampling distribution of x̄ is approximately Normal.
3
Step 3 — Identify Values and Find t*

x̄ = 2250, s = 420, n = 36, df = 36 − 1 = 35. For a 95% confidence interval, t* = 2.030 (from the t-table or calculator with df = 35).

t* = 2.030, df = 35
4
Step 4 — Calculate the Standard Error and Margin of Error

SE = s / √n = 420 / √36 = 420 / 6 = 70. ME = t* × SE = 2.030 × 70 = 142.1.

SE = 70, ME = 142.1
5
Step 5 — Construct the Confidence Interval

CI = x̄ ± ME = 2250 ± 142.1 = (2107.9, 2392.1). We are 95% confident that the true mean daily caloric intake μ is between 2107.9 and 2392.1 calories.

95% CI: (2107.9, 2392.1)
6
Step 6 — Evaluate the Claim

The nutritionist's claimed value μ₀ = 2100 falls outside the 95% confidence interval (2107.9, 2392.1). Because 2100 is not a plausible value for μ at the 95% confidence level, we have convincing evidence that the true mean daily caloric intake for college students at this university differs from 2100 calories. Equivalently, a two-sided test of H₀: μ = 2100 versus Hₐ: μ ≠ 2100 at α = 0.05 would reject H₀.

The data provide convincing evidence against the claim μ = 2100.

SECTION 7

Confidence Intervals vs. Hypothesis Tests: Strengths & Limitations

Confidence intervals and hypothesis tests are complementary approaches to inference, and the AP exam expects students to understand the relative advantages of each. While a hypothesis test delivers a yes/no decision, a confidence interval communicates the range of plausible values and the precision of the estimate, often providing richer information. However, each approach has specific limitations that are important to recognize in applied settings.

Feature	Confidence Interval	Hypothesis Test
Output	Range of plausible values for μ	Reject / fail to reject decision
Precision information	Directly shown by interval width (ME)	Not directly conveyed; requires separate effect size or CI
Directionality	Shows whether μ is likely above or below μ₀	Two-sided test does not specify direction unless one-sided
Applicability for one-sided claims	Standard two-sided CI only addresses two-sided alternatives directly	Can be tailored to one-sided alternatives
Duality	A C% CI and a two-sided test at α = 1 − C always agree	Equivalent for two-sided cases
Practical significance	Easier to assess—readers can judge if the range matters practically	Statistical significance ≠ practical significance; large n can make trivial differences significant

Comparison of confidence intervals and hypothesis tests for evaluating claims about μ

✦

KEY TAKEAWAY

Think of a hypothesis test as a courtroom verdict—guilty or not guilty—whereas a confidence interval is more like the full police report, providing the range of scenarios consistent with the evidence. Both use the same underlying data, but the interval gives a more nuanced picture. On the AP exam, when you are asked to justify a claim and a confidence interval is provided, use the interval approach: it directly answers whether the claimed value is plausible without requiring you to compute a test statistic separately.

SECTION 8

Connections to Advanced Inference

The logic of using a confidence interval to evaluate a claim about a single population mean extends naturally to more advanced inferential settings. Understanding these connections will prepare you for later topics in AP Statistics and for college-level statistics courses that introduce additional complexity.

This Lesson (One-Sample Mean)	Advanced Extension
CI for a single population mean μ	CI for the difference of two means μ₁ − μ₂ (two-sample t-interval): if 0 is not in the interval, there is evidence the means differ
Two-sided claim evaluation using CI	One-sided bounds (upper or lower confidence bounds) can evaluate one-sided claims directly
Fixed confidence level (e.g., 95%)	Adjusting for multiple comparisons (Bonferroni correction) when evaluating many claims simultaneously
t-interval with Normal assumption	Bootstrap confidence intervals that relax distributional assumptions, useful for skewed populations or small samples
Frequentist interpretation (long-run capture rate)	Bayesian credible intervals that assign a direct probability to the parameter being in the interval, given a prior distribution

How single-mean CI concepts generalize to advanced inference

For the AP Statistics exam, the most immediately relevant extension is the two-sample t-interval for a difference of means. The claim evaluation logic is identical: if the interval for μ₁ − μ₂ does not contain 0, there is convincing evidence that the two population means differ. Similarly, for matched-pairs designs, you construct a one-sample t-interval on the differences and check whether 0 is captured. In each case, the core reasoning—does the claimed value fall inside or outside the interval?—remains the same.

SECTION 9

Practice Problems

Question 1: CONCEPTUAL

A 90% confidence interval for the mean weight of packages shipped by a company is (15.3, 17.1) pounds. A manager claims the true mean weight is 16 pounds. Which of the following is the best conclusion based on this interval?

A. There is a 90% probability that the true mean weight is 16 pounds.
B. The data provide convincing evidence that the true mean weight is not 16 pounds.
C. The data do not provide convincing evidence that the true mean weight differs from 16 pounds.
D. Exactly 90% of all packages weigh between 15.3 and 17.1 pounds.
E. The claim μ = 16 has been proven true at the 90% confidence level.

Show answer (C)

C is correct. Because 16 falls inside the 90% confidence interval (15.3, 17.1), the data are consistent with the claim μ = 16; we do not have convincing evidence that the true mean differs from 16. A is wrong because the 90% refers to the method's long-run capture rate, not the probability that μ equals a specific value. B is wrong because 16 is inside the interval, so the data do not contradict this value. D is wrong because the confidence interval is about the population mean, not individual data values. E is wrong because a confidence interval can never prove a claim true; it can only show that the value is plausible.

Question 2: BASIC CALCULATION

A random sample of 25 light bulbs has a sample mean lifetime of 1200 hours and a sample standard deviation of 100 hours. A 95% confidence interval for the population mean lifetime is (1158.7, 1241.3). A manufacturer claims the mean lifetime is 1150 hours. Based on this interval, which of the following is correct?

A. The claim μ = 1150 is plausible because it is close to the lower bound of the interval.
B. The data provide convincing evidence that the true mean lifetime differs from 1150 hours at the α = 0.05 significance level.
C. The data provide convincing evidence that the true mean lifetime differs from 1150 hours at the α = 0.01 significance level.
D. The data do not provide convincing evidence that the true mean lifetime differs from 1150 hours.
E. A larger sample is needed before any conclusion can be drawn.

Show answer (B)

B is correct. The claimed value μ₀ = 1150 falls outside the 95% confidence interval (1158.7, 1241.3). By the duality between confidence intervals and hypothesis tests, this means a two-sided test at α = 0.05 would reject H₀: μ = 1150, so the data provide convincing evidence that the true mean differs from 1150 hours. A is wrong because 'close to the bound' does not matter—1150 is outside the interval. C is wrong because we cannot determine the α = 0.01 conclusion from a 95% CI; we would need a 99% CI. D is wrong because 1150 is outside the interval. E is wrong because the conditions are met and the interval has been constructed; a conclusion can be drawn.

Question 3: INTERMEDIATE

A researcher constructs both a 90% and a 99% confidence interval for the mean response time (in milliseconds) of a software system. The 90% CI is (245, 275) and the 99% CI is (235, 285). A competitor claims the true mean response time is 240 ms. Which of the following conclusions is supported by these intervals?

A. The data provide convincing evidence against the claim at both α = 0.10 and α = 0.01.
B. The data provide convincing evidence against the claim at α = 0.10 but not at α = 0.01.
C. The data provide convincing evidence against the claim at α = 0.01 but not at α = 0.10.
D. The data do not provide convincing evidence against the claim at either significance level.
E. No comparison can be made because the two intervals are based on different confidence levels.

Show answer (B)

B is correct. The claimed value μ₀ = 240 is outside the 90% CI (245, 275), so a two-sided test at α = 0.10 would reject H₀: μ = 240. However, 240 is inside the 99% CI (235, 285), so a two-sided test at α = 0.01 would fail to reject H₀. This illustrates how the choice of confidence level affects the conclusion. A is wrong because 240 is inside the 99% CI. C is wrong because it reverses the conclusions; rejecting at α = 0.01 (more stringent) without rejecting at α = 0.10 (less stringent) is logically impossible. D is wrong because 240 is outside the 90% CI. E is wrong because the whole point of constructing intervals at different levels is to assess claims at different significance thresholds.

Question 4: APPLIED

4 pts

A transportation engineer claims that the mean commute time for workers in a metropolitan area is 35 minutes. A city planner selects a simple random sample of 50 commuters and records their one-way commute times. The sample yields x̄ = 38.2 minutes and s = 12.5 minutes. (a) Construct a 95% confidence interval for the true mean commute time μ. Show your work. (b) Based on your interval, is there convincing evidence that the true mean commute time differs from 35 minutes? Justify your answer. (c) Without performing additional calculations, would a 99% confidence interval be more or less likely to contain 35? Explain. (d) What does the confidence interval tell you about commute times that a simple hypothesis test result (reject / fail to reject) would not?

Show answer

Rubric (4 points): Part (a) — 1 point: Conditions: Random (SRS stated), Independent (50 < 10% of all commuters in a metro area), Normal (n = 50 ≥ 30, CLT applies). df = 49, t* ≈ 2.010. SE = 12.5 / √50 ≈ 1.768. ME = 2.010 × 1.768 ≈ 3.554. CI = 38.2 ± 3.554 = (34.646, 41.754). Award 1 point for correct interval with conditions stated or verified. Part (b) — 1 point: The claimed value μ₀ = 35 falls inside the 95% CI (34.646, 41.754). Therefore, the data do not provide convincing evidence that the true mean commute time differs from 35 minutes at the α = 0.05 significance level. Award 1 point for correctly identifying 35 is inside the CI and drawing the appropriate conclusion with justification. Part (c) — 1 point: A 99% CI is wider than a 95% CI because the critical value t* increases with the confidence level. Since 35 is already inside the 95% CI, it would also be inside the wider 99% CI. Therefore, a 99% interval is more likely (in fact, certain in this case) to contain 35. Award 1 point for stating the 99% CI is wider and explaining that 35 would still be captured. Part (d) — 1 point: The confidence interval shows that the plausible values for μ range from about 34.6 to 41.8 minutes. This tells us both the direction and magnitude of the possible deviation from 35—the true mean could be as high as nearly 42 minutes. A simple reject/fail-to-reject result would only indicate whether 35 is plausible, without communicating the range of uncertainty or helping us assess practical significance. Award 1 point for explaining the additional information conveyed by the CI (range, direction, precision, or practical significance).

Question 5: CRITICAL THINKING

4 pts

A pharmaceutical company reports that a 95% confidence interval for the mean reduction in systolic blood pressure (mmHg) after taking a new medication is (1.2, 8.8). A rival company claims the medication has no effect (μ = 0). A health organization considers a reduction of at least 5 mmHg to be clinically meaningful. (a) Does the confidence interval provide convincing evidence that the medication reduces blood pressure at all? Justify using the interval. (b) Does the confidence interval provide convincing evidence that the mean reduction is at least 5 mmHg? Explain. (c) Discuss the distinction between statistical significance and practical (clinical) significance in the context of this problem. (d) A researcher proposes increasing the sample size substantially. Explain how this would affect the width of the confidence interval and, in turn, the conclusions from parts (a) and (b).

Show answer

Rubric (4 points): Part (a) — 1 point: Yes. The claimed value μ₀ = 0 (no effect) falls outside the 95% CI (1.2, 8.8). Because the entire interval lies above 0, there is convincing evidence at the α = 0.05 level that the medication does reduce systolic blood pressure. Award 1 point for identifying that 0 is outside the CI and stating the conclusion with justification. Part (b) — 1 point: No. The value 5 falls inside the 95% CI (1.2, 8.8), meaning the data are consistent with a mean reduction of 5 mmHg. However, the interval also includes values below 5 (such as 1.2, 2, 3, 4), so we cannot conclude that the mean reduction is at least 5 mmHg. The data do not provide convincing evidence that the mean reduction meets the clinically meaningful threshold. Award 1 point for identifying that 5 is inside the CI and explaining that values below 5 are also plausible. Part (c) — 1 point: Statistical significance means the data provide evidence that the true effect differs from zero (μ ≠ 0), which is established here because 0 is outside the CI. Practical (clinical) significance means the effect is large enough to matter in real-world medical practice—here, at least 5 mmHg. The interval shows the medication is statistically significant but may not be practically significant, since the true mean reduction could be as low as 1.2 mmHg. Award 1 point for clearly distinguishing statistical from practical significance and connecting both to the interval. Part (d) — 1 point: Increasing the sample size decreases the standard error (SE = s / √n), which in turn decreases the margin of error and produces a narrower confidence interval. The conclusion from part (a) would remain the same (0 would still be excluded), but a narrower interval could potentially resolve part (b): if the interval narrowed to, say, (4.0, 6.0), the lower bound would clarify whether the mean reduction is above or below 5. Conversely, if the interval narrowed to (2.5, 4.5), it would provide convincing evidence that the reduction does not reach 5 mmHg. Award 1 point for explaining the effect of larger n on ME/width and discussing implications for both conclusions.

SUMMARY

Lesson Summary

Summary

A confidence interval for a population mean μ provides a range of plausible values computed from sample data. To justify a claim about μ, check whether the claimed value μ₀ falls inside or outside the interval. If μ₀ is inside the interval, the data are consistent with the claim and do not provide convincing evidence against it. If μ₀ is outside the interval, the data provide convincing evidence that μ differs from the claimed value at the corresponding significance level α = 1 − C. This logic relies on the duality between a C% confidence interval and a two-sided hypothesis test at significance level α.

Key elements for a complete AP exam response include: stating the parameter and claimed value, verifying the conditions (random, independent, Normal), constructing or referencing the interval using the formula x̄ ± t* × (s / √n), and drawing a conclusion that uses correct language—saying "we are C% confident" rather than "there is a C% probability." Remember that the margin of error reflects the precision of the estimate and that a wider interval captures more values as plausible but provides less definitive conclusions about any single claim.