# Justifying a Claim Based on a Confidence Interval for the Difference Between Two Population Means

> AP Statistics · AP Stats 2024-2026
> Source: https://www.owlsprep.com/study/ap-statistics-u13-justifying-a-claim-based-on/

This module teaches you to use calculated two-sample confidence intervals for the difference of population means to formally justify statistical claims, aligned exactly with AP FRQ scoring requirements for inference responses.

**Prerequisites:** [Constructing confidence intervals for the difference of two population means](https://www.owlsprep.com/study/ap-statistics-u13-two-sample-t-interval-means/); [Foundations of two-sample t significance tests](https://www.owlsprep.com/study/ap-statistics-u13-two-sample-t-test-intro/)

## Learning objectives

- Interpret a two-sample mean confidence interval in context to support or refute a given statistical claim
- Identify if a hypothesized difference value falls inside or outside the calculated interval to draw a valid conclusion
- Explain the direct equivalence between two-tailed two-sample t-tests and corresponding confidence intervals
- Produce justification responses that meet full AP exam FRQ rubric requirements to avoid point deductions

## Core Logic of Claim Justification with Two-Mean Intervals

The core rule for justifying a claim using a two-sample confidence interval for $\mu_1 - \mu_2$ is extremely straightforward: any value that lies inside the interval is a plausible value for the true population difference, and any value that lies outside the interval is not a plausible value at the corresponding confidence level.

**Null Value Check** — To evaluate a claim about a hypothesized difference $d_0$, you only need to check if $d_0$ falls within the bounds of your calculated confidence interval.

*Notation:* $H_0: \mu_1 - \mu_2 = d_0$

*Example:* If your 95% interval for the difference in test scores between two classes is (-2.3, 4.1), the null value 0 is inside the interval, so no significant difference exists at $\alpha=0.05$.

**Worked example:** A 95% confidence interval for the difference in mean daily screen time (teenagers minus adults) is calculated as (1.2 hours, 2.9 hours). Justify the claim that teenagers have a different mean daily screen time than adults.

1. Step 1: Identify the hypothesized difference value for the claim of no difference.
2. $$d_0 = 0$$
3. Step 2: Check if 0 falls inside the given interval (1.2, 2.9).
4. Step 3: 0 is less than the lower bound of 1.2, so it is not a plausible value for the true difference.
5. Step 4: Conclusion: We have statistically significant evidence at the 95% confidence level that the mean screen time for teenagers is different from adults.

**Check your understanding**

Test your basic understanding of the null value check rule:

1. A 95% interval for $\mu_1 - \mu_2$ is (-3.2, -0.7). Is 0 a plausible value for the true difference?

   - Yes
   - No

   *Why:* 0 is greater than the upper bound of -0.7, so it is outside the interval and not a plausible value.

## AP Exam Rubric Aligned Justification Structure

**Exam command terms**

AP exam readers award full points only if you use the exact required phrasing structure for justification, no shortcuts are accepted:

- **"The interval contains / does not contain"** — Explicitly state the position of the hypothesized difference value relative to the interval bounds, do not skip this step *("Our 95% confidence interval for the difference in mean plant growth (fertilizer minus control) is (0.4 cm, 2.1 cm), and it does not contain the null value 0.")*

- **"We have statistically significant evidence"** — Qualify your conclusion to avoid overstating results, and always reference the population parameters not sample statistics *("We have statistically significant evidence at the 0.05 level that the fertilizer increases mean plant growth.")*

> **Rubric Pro Tip**
>
> You will lose the 'complete response' point if you only state your final conclusion without explicitly referencing the interval bounds and the hypothesized difference value.

**Worked example:** A researcher calculates a 90% confidence interval for the difference in mean commute time (Route A minus Route B) as (-1.1 minutes, 3.4 minutes). Justify the claim that Route A has a different mean commute time than Route B, following AP rubric rules.

1. Step 1: State the hypothesized difference value for the no-difference claim: 0.
2. Step 2: Explicitly note that 0 falls inside the interval bounds of -1.1 and 3.4 minutes.
3. Step 3: Confirm that 0 is a plausible value for the true difference in population mean commute times.
4. Step 4: Final conclusion: We do not have statistically significant evidence at the 90% confidence level that Route A and Route B have different mean commute times.

## Equivalence Between Intervals and Two-Tailed t-Tests

**Derivation:** Show that a 100(1-α)% confidence interval for $\mu_1 - \mu_2$ is exactly equivalent to a two-tailed two-sample t-test with significance level $\alpha$

*Starting from:* The formula for a two-sample t confidence interval: $\bar{x}_1 - \bar{x}_2 \pm t^* \times SE(\bar{x}_1 - \bar{x}_2)$

1. 1. A two-tailed t-test rejects $H_0$ when the t test statistic is more extreme than the critical value $t^*$.
2. 2. Rearranging the t-test inequality shows that the null value $d_0$ will fall outside the interval exactly when the test rejects $H_0$.
3. 3. The confidence level of 1-α matches exactly to the two-tailed significance level α.

*Conclusion:* A 95% confidence interval corresponds perfectly to a two-tailed t-test with $\alpha=0.05$.

**Comparing methods**

Compare the two valid methods for justifying a claim about two population means:

- **Confidence Interval Justification** — Check if the hypothesized difference falls inside the interval bounds
  - Pros: Requires no additional calculations, directly uses interval results you already computed
  - Cons: Only valid for two-tailed tests, cannot be used for one-sided significance checks

- **Two-Sample t-Test Justification** — Calculate t test statistic and p-value to compare to significance level
  - Pros: Works for both one-tailed and two-tailed tests
  - Cons: Requires extra calculation steps that are unnecessary if you already have an interval

**Worked example:** A 99% confidence interval for the difference in mean exam scores (class 1 minus class 2) is (-5.2, 1.3). What conclusion would a two-tailed t-test for no difference at $\alpha=0.01$ produce?

1. Step 1: Confirm that 99% confidence level corresponds exactly to two-tailed $\alpha=0.01$.
2. Step 2: Check if the null value 0 falls inside the interval (-5.2, 1.3).
3. Step 3: 0 is inside the interval, so the t-test will fail to reject the null hypothesis of no difference.
4. Step 4: Final conclusion: There is no statistically significant difference in mean exam scores between the two classes at the 0.01 level.

## Justifying One-Sided Directional Claims

> **Critical Warning**
>
> Never use a two-sided confidence interval to justify a one-tailed test at the same alpha level, this will produce an incorrect p-value and invalid conclusion.

If your entire two-sided confidence interval for $\mu_1 - \mu_2$ is positive, you can justify the one-sided claim that $\mu_1 > \mu_2$ at a significance level of $\alpha/2$. If the entire interval is negative, you can justify the one-sided claim that $\mu_1 < \mu_2$ at a significance level of $\alpha/2$.

**Worked example:** A 95% two-sided confidence interval for the difference in mean weight loss (diet plan X minus diet plan Y) is (0.8 kg, 3.2 kg). Justify the claim that diet plan X produces greater mean weight loss than diet plan Y.

1. Step 1: Confirm the interval is entirely positive, with all plausible values for $\mu_X - \mu_Y$ greater than 0.
2. Step 2: The corresponding one-sided significance level is 0.05 / 2 = 0.025.
3. Step 3: Since 0 is less than the lower bound of the interval, we have statistically significant evidence at $\alpha=0.025$ that diet plan X leads to greater mean weight loss than diet plan Y.

## Common pitfalls

- **Wrong:** Stating the difference in sample means is inside the interval to justify a claim
  - Why it fails: The interval estimates the unknown population difference, not the already known sample difference, and this error earns zero points on AP rubrics
  - Correct: Explicitly reference the hypothesized population difference value relative to the interval bounds
- **Wrong:** Claiming there is a 95% probability the true difference is positive from an entirely positive interval
  - Why it fails: Confidence level describes the long-run capture rate of the interval method, not a probability for a single calculated interval
  - Correct: State "we are 95% confident the true difference in population means is positive" to avoid invalid probabilistic language
- **Wrong:** Using a 90% confidence interval to justify a conclusion for a two-tailed $\alpha=0.05$ hypothesis test
  - Why it fails: A 90% interval corresponds to $\alpha=0.10$, so the significance thresholds do not match and will produce inconsistent conclusions
  - Correct: Set the confidence level equal to $1-\alpha$ for the two-tailed test you are aligning to
- **Wrong:** Ignoring the order of subtraction ($\mu_1 - \mu_2$ vs $\mu_2 - \mu_1$) when justifying a claim
  - Why it fails: Flipping the order of subtraction reverses the sign of all interval bounds, leading to the exact opposite conclusion
  - Correct: Explicitly restate which group is defined as group 1 and group 2 before referencing interval bounds
- **Wrong:** Claiming the confidence interval proves one population mean is larger than the other with 100% certainty
  - Why it fails: All confidence intervals have a non-zero failure rate equal to the significance level, so absolute proof is impossible
  - Correct: Qualify all conclusions with the stated confidence level, noting the interval method will fail to capture the true difference at the expected rate

## Cheatsheet

| Scenario | Hypothesized $\mu_1 - \mu_2$ | 95% CI for $\mu_1 - \mu_2$ | Conclusion for $\alpha=0.05$ Two-Tailed Test |
| --- | --- | --- | --- |
| Claim: No difference between groups | 0 | (-2.1, 1.7) | Fail to reject $H_0$: No significant difference |
| Claim: $\mu_1 > \mu_2$ | 0 | (1.2, 4.8) | Reject $H_0$: Significant evidence $\mu_1 > \mu_2$ |
| Claim: $\mu_1 < \mu_2$ | 0 | (-5.3, -0.9) | Reject $H_0$: Significant evidence $\mu_1 < \mu_2$ |
| Claim: Difference of at least 3 units | 3 | (0.8, 2.7) | Fail to support claim: 3 is outside interval |

## What's next

Mastering this skill is critical for earning full inference points on the AP Statistics FRQ section, as nearly every two-sample means question requires you to connect interval results to a real-world research claim. This content directly builds on your prior work constructing two-sample t intervals and running two-sample t hypothesis tests, and it will prepare you to tackle more complex inference scenarios including paired t procedures and inference for linear slope. You will now be able to avoid the most common rubric point deductions that trip up even top-performing students on exam day. Practice applying these justification rules to full FRQ prompts to reinforce your understanding.

---

From [OwlsPrep](https://www.owlsprep.com) — free study guides for A-Level, IB, AP and IGCSE, written against the official syllabus. Canonical page: https://www.owlsprep.com/study/ap-statistics-u13-justifying-a-claim-based-on/