Welcome to Comparing Two Groups!

In our previous chapters, we looked at how to estimate a single population mean. But in the real world, we often want to compare two different groups. For example: Do students who use a specific app study more hours than those who don't? or Is there a difference in the average lifespan of two different brands of lightbulbs?

In this chapter, we will learn how to build a Confidence Interval for the Difference Between Two Population Means \((\mu_1 - \mu_2)\). This helps us estimate how much of a difference actually exists between two independent groups. Don't worry if this seems a bit more complex than the one-sample version—the logic is exactly the same, just with a few more pieces to keep track of!


1. The Big Idea: The Point Estimate

When we want to know the difference between two population means \((\mu_1 - \mu_2)\), we start with what we know: our sample means. Our point estimate is simply the difference between our two sample means:

Point Estimate = \(\bar{x}_1 - \bar{x}_2\)

Analogy: If you want to estimate the difference in height between basketball players and soccer players, you'd take a sample of each, find their average heights, and subtract them. That difference is your starting point!


2. The General Formula

All confidence intervals in AP Statistics follow the same "skeleton" formula:

Statistic \(\pm\) (Critical Value) \(\times\) (Standard Error of the Statistic)

For the difference between two means, it looks like this:

\((\bar{x}_1 - \bar{x}_2) \pm t^* \sqrt{\frac{s_1^2}{n_1} + \frac{s_2^2}{n_2}}\)

Breaking down the parts:
  • \(\bar{x}_1 - \bar{x}_2\): The difference between your two sample means (your "best guess").
  • \(t^*\): The critical value, which depends on your confidence level and degrees of freedom (df).
  • \(\sqrt{\frac{s_1^2}{n_1} + \frac{s_2^2}{n_2}}\): This is the Standard Error (SE). It accounts for the variability in both samples.

Quick Note on Degrees of Freedom: Calculating degrees of freedom for two samples by hand is very messy! The AP exam (and the Fall 2026 CED) expects you to use technology (like your graphing calculator) to find the exact degrees of freedom. If you ever have to do it without a calculator, a safe "conservative" estimate is to use the smaller of \(n_1 - 1\) or \(n_2 - 1\).


3. Conditions for Inference

Before we can build our interval, we must verify that our data is "healthy" enough to use. Never skip this step on a Free-Response Question!

  1. Random: Data must come from two independent random samples or a randomized experiment. This allows us to generalize to the populations.
  2. 10% Condition: If we are sampling without replacement, our sample sizes must be less than 10% of their respective populations (\(n_1 < 10\% N_1\) and \(n_2 < 10\% N_2\)). Note: We don't check this for randomized experiments!
  3. Normal/Large Sample: For each group, we need to ensure the sampling distribution of the mean is approximately Normal. We check this by:
    • The population is stated to be Normal.
    • OR the sample size is large (\(n \ge 30\)). This is the Central Limit Theorem at work!
    • OR if the sample is small (\(n < 30\)), graph the sample data. As long as there is no strong skewness or extreme outliers, we are good to go.

Common Mistake: Students often forget to check the conditions for both groups. You must show that Group 1 and Group 2 meet the requirements!


4. Step-by-Step Construction

When you are asked to "Construct and interpret a confidence interval," follow these steps:

Step 1: Identify the Parameters

Clearly state what \(\mu_1\) and \(\mu_2\) represent in the context of the problem.
Example: \(\mu_1\) = the true mean test score for students using the new textbook.

Step 2: Check Conditions

Verify Random, 10%, and Normal/Large Sample for both groups (as shown above).

Step 3: Calculate

Use your calculator's 2-Sample T-Interval function. You will need the mean (\(\bar{x}\)), standard deviation (\(s\)), and sample size (\(n\)) for both groups.
Important: If your calculator asks about "Pooling," always select No. We do not pool variances for means in AP Statistics.

Step 4: Interpret

Write your conclusion using this template:
"We are [95]% confident that the interval from [lower bound] to [upper bound] captures the true difference in mean [context] between [Group 1] and [Group 2]."


5. Justifying a Claim

One of the most important skills in Unit 4 is using your interval to make a decision. Usually, we want to know if there is a convincing difference between the two groups.

The "Zero" Rule:

  • If zero (0) is NOT in the interval (e.g., the interval is all positive numbers like [2, 5]), we have convincing evidence of a difference between the two population means.
  • If zero (0) IS in the interval (e.g., the interval is [-3, 4]), we do not have convincing evidence of a difference. It’s possible the true difference is zero!

Did you know? If your interval for \((\mu_1 - \mu_2)\) is entirely positive, it suggests that \(\mu_1\) is likely larger than \(\mu_2\). if it is entirely negative, \(\mu_2\) is likely larger!


Key Takeaways

Summary Checklist:

  • The interval estimates the difference between two independent population means.
  • Always check conditions for both samples.
  • Use technology for Degrees of Freedom and to calculate the interval.
  • If 0 is in the interval, there is no convincing evidence of a difference.
  • Never "pool" your data for means.

Quick Review: Remember that this is different from a Paired T-Interval (from Chapter 4.2). We use 2-Sample T-Intervals when the groups are independent (like Boys vs. Girls). We use Paired T-Intervals when the same subjects are measured twice (like Before vs. After) or are naturally matched.