Quick Summary
This guide will equip you to use a confidence interval for the difference between two population means (μ₁ - μ₂) to make a formal conclusion about a claim. You will learn how to determine if there is a statistically significant difference between two group means by checking whether the value of zero is contained within the interval. This skill allows you to use the richer information from a confidence interval to perform the equivalent of a two-sided hypothesis test.
Key Concepts
The core idea of this topic is that a confidence interval provides a range of plausible values for a population parameter. When our parameter is the difference between two population means (μ₁ - μ₂), the interval gives us a range of plausible values for that true difference. We can leverage this to test the claim that there is no difference, which corresponds to a null hypothesis of H₀: μ₁ - μ₂ = 0.
The "Is Zero In There?" Rule
The decision-making process is simple and powerful. After a confidence interval for μ₁ - μ₂ has been calculated, you only need to ask one question: Is the value 0 contained within the interval?
Case 1: Zero IS in the confidence interval.
Example: A 95% confidence interval for the difference in mean commute times (μ_Car - μ_Train) is (-3.5 minutes, 6.2 minutes).
Interpretation: Because 0 is a value within this interval, it is a plausible value for the true difference in means. This means it's plausible that there is no difference at all between the true mean commute times.
Conclusion: We do not have convincing statistical evidence of a difference between the two population means. This is the same conclusion we would reach if we failed to reject the null hypothesis (H₀: μ₁ - μ₂ = 0) in a two-sided significance test at the corresponding alpha level (α = 1 - Confidence Level).
Case 2: Zero IS NOT in the confidence interval.
Example: A 99% confidence interval for the difference in mean crop yield (μ_FertilizerA - μ_FertilizerB) is (2.1 bushels, 8.7 bushels).
Interpretation: Because 0 is not a value within this interval, it is not a plausible value for the true difference in means. All the plausible values are positive.
Conclusion: We have convincing statistical evidence of a difference between the two population means. This is the same conclusion we would reach if we rejected the null hypothesis (H₀: μ₁ - μ₂ = 0).
Determining the Direction of the Difference
When the interval does not contain zero, we can go a step further and describe the direction of the difference.
If all values in the interval are positive:
Example: (2.1, 8.7)
Meaning: All plausible values for μ₁ - μ₂ are positive. This implies that μ₁ - μ₂ > 0, which means μ₁ > μ₂.
Conclusion: We have convincing evidence that the true mean for the first group is greater than the true mean for the second group.
If all values in the interval are negative:
Example: (-10.4, -1.9)
Meaning: All plausible values for μ₁ - μ₂ are negative. This implies that μ₁ - μ₂ < 0, which means μ₁ < μ₂.
Conclusion: We have convincing evidence that the true mean for the first group is less than the true mean for the second group.
[Image: A number line showing three confidence intervals relative to zero. The first interval, labeled "Evidence μ₁ < μ₂," is entirely to the left of zero (e.g., -10 to -2). The second interval, labeled "No evidence of a difference," crosses zero (e.g., -5 to 5). The third interval, labeled "Evidence μ₁ > μ₂," is entirely to the right of zero (e.g., 2 to 10).]
Connecting Confidence Level to Significance Level
Using a confidence interval to justify a claim is equivalent to performing a two-sided significance test. The significance level (α) of the test is related to the confidence level (C) of the interval by the formula: α = 1 - C.
A 90% confidence interval corresponds to a test at the α = 0.10 significance level.
A 95% confidence interval corresponds to a test at the α = 0.05 significance level.
A 99% confidence interval corresponds to a test at the α = 0.01 significance level.
This connection is critical for making conclusions at a specific significance level.
Key Vocabulary
Confidence Interval: An interval of plausible values for an unknown population parameter, calculated from sample data.
Difference of Two Means (μ₁ - μ₂): The parameter of interest. It represents the true, unknown difference between the average values of two distinct populations.
Statistically Significant: An observed result is statistically significant if it is too unlikely to have occurred by random chance alone, assuming the null hypothesis is true. In this context, a difference is significant if its confidence interval does not contain zero.
Plausible Value: Any value for a parameter that is contained within a confidence interval. These are the values that are not contradicted by the sample data at the given confidence level.
Null Hypothesis (H₀): A statement of "no effect" or "no difference." For the difference of two means, the null hypothesis is almost always H₀: μ₁ - μ₂ = 0, which states that the two population means are equal.
Calculator Tech (TI-84)
This topic focuses on interpreting a pre-existing confidence interval, so no calculations are typically required. However, if you were to calculate the interval yourself from data or summary statistics, you would use the function.
Path:STAT -> TESTS -> 0: 2-SampTInt...
You would then choose (if you have raw data in lists L1 and L2) or (if you have summary statistics like x̄, s, and n for both groups) and enter the required values and the confidence level (C-Level). The output provides the confidence interval . For Topic 7.7, this interval will usually be given to you in the problem prompt.
How to Show Work on the FRQ
When an FRQ asks you to justify a claim about the difference of two means based on a given confidence interval, a full State-Plan-Do-Conclude (SPDC) is not necessary. Your response should be a concise, two-part conclusion.
FRQ Template for Interpreting a Confidence Interval
Part 1: The Decision Rule (The "Zero" Statement)
State clearly whether the value 0 is contained within the given confidence interval.
- Sentence Frame: "The [C-level]% confidence interval for the difference in true mean [context] (Group 1 - Group 2) is ([lower bound], [upper bound]). The value 0 is / is not contained in this interval."
Part 2: The Conclusion in Context
Based on your statement in Part 1, write a conclusion that directly answers the question asked in the prompt. Use appropriate statistical language and maintain context.
Scenario A: 0 IS in the interval.
- Sentence Frame: "Because 0 is a plausible value for the true difference in means, we do not have convincing statistical evidence to conclude that there is a difference between the true mean [context for Group 1] and the true mean [context for Group 2]."
Scenario B: 0 IS NOT in the interval.
Sentence Frame: "Because 0 is not a plausible value for the true difference in means, we have convincing statistical evidence to conclude that there is a difference between the true mean [context for Group 1] and the true mean [context for Group 2]."
Add Direction (if applicable): "Furthermore, since all the values in the interval are [positive/negative], we have evidence that the true mean [context for Group 1] is greater than / less than the true mean [context for Group 2]."
Practice Problems
Problem 1:
A school administrator wants to know if there is a difference in the mean number of hours that junior and senior students spend on homework each week. A random sample of 20 juniors and a separate random sample of 25 seniors are selected. A 95% confidence interval for the difference in the true mean number of homework hours (μ_juniors - μ_seniors) was calculated to be (-1.8 hours, 2.5 hours). Based on this interval, is there convincing evidence of a difference in the mean homework hours between juniors and seniors at this school?
Solution:
Part 1: The Decision Rule
The 95% confidence interval for the difference in true mean homework hours (juniors - seniors) is (-1.8, 2.5). The value 0 is contained in this interval.
Part 2: The Conclusion in Context
Because 0 is a plausible value for the true difference in means, we do not have convincing statistical evidence to conclude that there is a difference between the true mean number of homework hours for juniors and the true mean number of homework hours for seniors at this school.
Problem 2:
A pharmaceutical company develops a new drug designed to lower blood pressure. To test its effectiveness, they recruit 50 volunteers. They randomly assign 25 volunteers to receive the new drug and 25 to receive a placebo. After six weeks, the reduction in systolic blood pressure is measured for each volunteer. A 99% confidence interval for the difference in mean blood pressure reduction (μ_drug - μ_placebo) is (4.2 mmHg, 9.8 mmHg). Based on this interval, is there convincing evidence that the new drug is more effective than the placebo at reducing blood pressure?
Solution:
Part 1: The Decision Rule
The 99% confidence interval for the difference in true mean blood pressure reduction (drug - placebo) is (4.2, 9.8). The value 0 is not contained in this interval.
Part 2: The Conclusion in Context
Because 0 is not a plausible value for the true difference in means, we have convincing statistical evidence to conclude that there is a difference between the true mean blood pressure reduction for the drug and the true mean blood pressure reduction for the placebo. Furthermore, since all the values in the interval are positive, we have convincing evidence that the true mean blood pressure reduction for the drug group is greater than the true mean blood pressure reduction for the placebo group. This suggests the drug is effective.
Common Mistakes to Avoid
Concluding About Samples, Not Populations: Your conclusion must always be about the population means (μ₁ and μ₂), not the sample means (x̄₁ and x̄₂). The entire purpose of inference is to use sample data to draw conclusions about the larger populations.
Incorrect: "There is no evidence of a difference in the sample mean homework hours."
Correct: "There is no evidence of a difference in the true mean homework hours."
"Accepting the Null Hypothesis": When the interval contains 0, you must state that you "do not have convincing evidence of a difference" or that you "fail to reject the null hypothesis." You must never state that you "accept the null hypothesis" or that the means "are equal." A lack of evidence against a claim is not the same as proof that the claim is true.
Forgetting Context: Always phrase your conclusion in the context of the problem. Do not just say "we have evidence of a difference." State what you are measuring—blood pressure, test scores, homework hours, etc.
Misinterpreting the Order of Subtraction: Pay close attention to how the difference is defined (e.g., Group A - Group B). If the interval is entirely positive, it means Group A's mean is greater. If it's entirely negative, it means Group B's mean is greater. A common mistake is to see a negative interval like (-10, -2) and incorrectly conclude Group A's mean is smaller, but forget to state that this means Group B's mean is larger.
Using Language of Certainty: Avoid absolute words like "prove" or "confirm." Statistical inference provides evidence, not proof. Stick to phrases like "we have convincing evidence" or "the data suggest."