Welcome to Unit 6: Making Sense of Proportions!

In the previous units, we learned how to collect data and describe it. Now, we are entering the heart of statistics: Inference. Inference is just a fancy word for "making an educated guess about a whole population based on a small sample."

In Unit 6, we focus specifically on categorical data. This is data that falls into categories (Yes/No, Red/Blue, Pass/Fail). We use proportions (percentages) to describe this data. By the end of this unit, you’ll be able to claim things like, "I am 95% confident that between 60% and 70% of students prefer pizza over tacos." Let’s dive in!

1. The Foundation: Confidence Intervals for One Proportion

Imagine you want to know what percentage of all teenagers in the U.S. use a specific app. You can’t ask every single teen, so you ask a random sample of 100. If 60 of them say "Yes," your point estimate (\(\hat{p}\)) is 0.60. But is the true population proportion exactly 0.60? Probably not. We use a Confidence Interval to give us a range of plausible values.

The "Big Three" Conditions

Before we can do any math, we must check three conditions. If these aren't met, our results might be "trash." Think of the mnemonic RIN:

• Random: The data must come from a random sample or randomized experiment. This prevents bias.
• Independent (10% Rule): When sampling without replacement, your sample size \(n\) must be less than 10% of the total population. This ensures the math stays consistent.
• Normal (Large Counts): You need at least 10 "successes" and 10 "failures." Mathematically: \(n\hat{p} \geq 10\) and \(n(1-\hat{p}) \geq 10\). This ensures our sampling distribution looks like a bell curve.

The Formula

A confidence interval always looks like this: Statistic ± Margin of Error.
For one proportion, the formula is: \(\hat{p} \pm z^* \sqrt{\frac{\hat{p}(1-\hat{p})}{n}}\)

Quick Tip: \(z^*\) is the "critical value." For a 95% confidence level, \(z^*\) is 1.96. You can find this on your formula sheet or using a calculator!

Key Takeaway: A confidence interval doesn't tell us where the data is; it tells us where we think the true population parameter is hiding.

2. Hypothesis Testing: The "Is It Weird?" Test

Hypothesis testing is like a court trial. We start by assuming the "status quo" is true (The Null Hypothesis, \(H_0\)). We then look at our sample data to see if it’s so weird that we have to reject that assumption in favor of a new claim (The Alternative Hypothesis, \(H_a\)).

The Steps of a Significance Test

1. State: Define your hypotheses. \(H_0: p = p_0\) and \(H_a: p > p_0\) (or \(<\) or \(\neq\)).
2. Plan: Identify the method (One-sample z-test for \(p\)) and check your RIN conditions.
3. Do: Calculate the test statistic (\(z\)) and the p-value.
4. Conclude: Compare your p-value to your significance level (\(\alpha\), usually 0.05).

The P-Value: The Most Important Number

The p-value is the probability of getting our sample result if the null hypothesis is actually true.
• If the P is low, the Null must go! (If \(p < \alpha\), we reject \(H_0\)).
• If the P is high, the Null can fly! (If \(p \geq \alpha\), we fail to reject \(H_0\)).

Analogy: Imagine a friend claims they are a 90% free-throw shooter. They take 10 shots and miss every single one. The probability of that happening by pure luck is tiny (low p-value). You would reject their claim!

Key Takeaway: We never "prove" the null is true. We only "fail to reject" it, meaning we don't have enough evidence to call it a lie yet.

3. Comparing Two Proportions

Sometimes we want to compare two different groups. For example: "Is the proportion of people who support a law higher in California than in Texas?"

Two-Sample z-Interval

This tells us the difference between two proportions (\(p_1 - p_2\)).
Conditions: You check the RIN conditions for both samples separately. Also, the two samples must be independent of each other.

Two-Sample z-Test and "Pooling"

When testing if two proportions are equal (\(H_0: p_1 = p_2\)), we use a special trick called pooling. Since we assume the two proportions are the same, we combine the samples to get one big "pooled" proportion (\(\hat{p}_c\)).
Formula for Pooled Proportion: \(\hat{p}_c = \frac{X_1 + X_2}{n_1 + n_2}\) (Total successes / Total sample size).

Don't worry if this seems tricky at first: Just remember that pooling is only for two-sample tests, never for intervals!

Key Takeaway: If a 95% confidence interval for the difference (\(p_1 - p_2\)) contains zero, it means there is no significant difference between the two groups.

4. Common Mistakes to Avoid

• Using \(\hat{p}\) instead of \(p\) in Hypotheses: Hypotheses are always about the population (\(p\)), never the sample (\(\hat{p}\)). You already know what the sample did!
• Interpreting Confidence Intervals Wrong: Do NOT say "There is a 95% chance the proportion is in this interval." Instead, say "We are 95% confident that the interval from [A] to [B] captures the true population proportion."
• Forgetting the 10% Condition: Students often skip this. If you are sampling from a finite group (like "all students at my school"), you must check it!

5. Quick Review Box

• Standard Error: The estimated standard deviation of our sample proportion. It gets smaller as the sample size \(n\) gets bigger.
• Margin of Error: Half the width of your confidence interval. To make it smaller (more precise), increase your sample size.
• Statistical Significance: When the p-value is smaller than \(\alpha\). It means the result was unlikely to happen by random chance alone.
• Did you know? Increasing confidence (e.g., from 90% to 99%) makes your interval wider. Think of it like a net: to be more sure you catch the fish, you need a bigger net!

You've got this! Unit 6 is all about the logic of the "If/Then." If the world works a certain way, then how likely is the sample I just saw? Keep practicing those "State, Plan, Do, Conclude" steps, and the patterns will start to feel like second nature.