Welcome to the "Toolbox" of Statistics!

In AP Statistics, you spend a lot of time learning how to calculate \( p \)-values and confidence intervals. But on the exam, the biggest challenge isn't the math—it's deciding which method to use in the first place! Think of this chapter as your guide to picking the right tool for the job. Whether you are dealing with percentages, averages, or categories, choosing the correct inference method is the "Practice 2.C" and "Practice 4.E" skill that ensures your conclusions are valid.

Don't worry if this seems overwhelming at first. We are going to break it down into a simple "decision tree" that works every time!

Step 1: Identify Your Variable Type

Before you can pick a test, you must know what kind of data you have. Ask yourself: "What are we measuring?"

  • Categorical Data (Proportions): If the data falls into categories (like "Yes/No," "Success/Failure," or "Color"), you are working with proportions. Look for the symbol \( p \) or \( \hat{p} \). (Unit 3)
  • Quantitative Data (Means): If the data is numerical and you can calculate an average (like "Height," "Score," or "Time"), you are working with means. Look for the symbol \( \mu \) or \( \bar{x} \). (Unit 4)

Step 2: Choose the Specific Inference Method

Once you know if you have proportions or means, count your samples and identify your goal.

Methods for Proportions (Categorical)

  • One-Sample \( z \)-Interval/Test: Used for a single group compared to a claim (e.g., "Is more than \( 50\% \) of the school voting for Taylor?").
  • Two-Sample \( z \)-Interval/Test: Used to compare two distinct groups (e.g., "Is the proportion of boys who like pizza different from the proportion of girls?").
  • Chi-Square Test for Homogeneity: Used when you have one categorical variable measured across two or more independent groups to see if the distributions are the same.
  • Chi-Square Test for Independence: Used when you have two categorical variables measured from one single sample to see if there is an association between them.

Methods for Means (Quantitative)

  • One-Sample \( t \)-Interval/Test: Used for one group compared to a claimed average (e.g., "Is the average weight of a cereal box really \( 16 \) ounces?").
  • Paired \( t \)-Interval/Test: This is a special type of one-sample test. It is used when you have two measurements for the same individual (like "Before" and "After") or matched pairs. You are testing the mean difference \( \mu_d \).
  • Two-Sample \( t \)-Interval/Test: Used to compare the averages of two independent groups (e.g., "Do students who study with music score higher on average than those who study in silence?").

Quick Note: You might have noticed we use \( z \) for proportions and \( t \) for means. A simple way to remember: "Mean-y Mr. \( T \)" (Means use \( t \)).

Step 3: Verifying Conditions (The "Safety Check")

Before you perform the math, you must prove that the data is "safe" to use. In AP Statistics, we call this verifying conditions. If these aren't met, your results might be biased or incorrect!

1. The Randomization Condition

What to look for: Did the data come from a random sample or a randomized experiment?

  • Why? It allows us to generalize our results to the population (for samples) or draw cause-and-effect conclusions (for experiments).
  • What to write: "It is stated that the sample was randomly selected," or "Treatments were randomly assigned."

2. The Independence (10%) Condition

What to look for: Is your sample size \( n \) less than \( 10\% \) of the total population \( N \)?

  • Why? When we sample without replacement, the observations aren't perfectly independent. If the sample is small enough (under \( 10\% \)), the math still works fine.
  • The Formula: \( n \le 0.10N \)
  • Note: You only check this if you are sampling without replacement. Do not check this for randomized experiments!

3. The Normality / Large Sample Condition

This condition ensures the sampling distribution is approximately Normal so we can use our \( z \) and \( t \) tables.

For Proportions: "Large Counts"

You must have at least \( 10 \) expected successes and \( 10 \) expected failures.

  • \( np \ge 10 \) and \( n(1-p) \ge 10 \) (Use \( \hat{p} \) for intervals).
For Means: "The Central Limit Theorem (CLT) or Normal Population"

Any one of these three will work:

  1. The population is stated to be normally distributed.
  2. The sample size is large enough (\( n \ge 30 \)). This is the Central Limit Theorem.
  3. If \( n < 30 \), look at a graph of the sample data. If there are no outliers and no strong skewness, it is safe to proceed.
For Chi-Square: "Expected Counts"
  • All expected counts must be at least \( 5 \). (Check the table of expected values, not the observed ones!)

Common Pitfalls to Avoid

1. Confusion between Paired \( t \) and Two-Sample \( t \):

If you have two lists of numbers, ask: "Are these linked?" If the first number in List A must be paired with the first number in List B (like a husband and wife, or the same person's left hand and right hand), it's Paired. If they are just two random groups (like Group A and Group B), it's Two-Sample.

2. Generic Condition Checklists:

The exam graders hate "generic" checks. Don't just write "Normal." Write "Since \( n = 45 \), which is greater than \( 30 \), the sampling distribution of the mean is approximately normal due to the Central Limit Theorem."

3. Forgetting the Context:

Always link your conclusion back to the story. Instead of saying "We reject the null," say "We have convincing evidence that the true proportion of students who eat breakfast is higher than \( 0.60 \)."

Key Takeaway Summary

Choosing the method is a three-step process:

  1. Data Type: Is it categorical (proportions) or quantitative (means)?
  2. Goal: Are you estimating a value (interval) or testing a claim (test)?
  3. Conditions: Did you check Random, 10%, and Normal/Large Counts?

Did you know? The Fall 2026 CED removed inference for the slope of a regression line! This means you can focus all your inference energy on Proportions, Means, and Chi-Square. One less tool to worry about in your toolbox!

For more details on how to actually calculate these results, cross-reference the chapters on "Carrying out a test" in Units 3 and 4.