Unit 6 Part 1: Comprehensive Notes on Confidence Intervals for Proportions

Overview of Statistical Inference for Proportions

Definition of Inference: Inference is the process of using sample statistics to make judgments, predictions, or estimates about a larger population parameter.
Context of Proportions: Proportions are primarily derived from categorical data (e.g., the proportion of students who completed homework, or the proportion of males vs. females who did so).
The Two Procedures of Inference: * Confidence Intervals: Used to estimate a population parameter based on a sample statistic. It provides a range of plausible values for the parameter. * Significance Tests: Used to determine the validity of a claim made about a population parameter (i.e., is the claim likely true or false?).
Scope of Application: These procedures can be applied to a single sample from one population or to the difference between two samples from two different populations.

Necessary Conditions for Inference

Before performing any inference procedures for proportions, three conditions must be met to ensure the validity of the results:

Randomness: The sample must be selected randomly to avoid bias. This ensures the sample is representative of the population.
Independence (The 10% Rule): The sample size $n$ must be less than 10% of the total population ( $N$ ). This allows researchers to assume independence between observations when sampling without replacement.
Large Counts (Shape/Normality): The sample size must be large enough so that the sampling distribution of the sample proportion is approximately normal. For proportions, this is satisfied if there are at least 10 successes and 10 failures in the sample: * $np ext{ (or } n ext{p-hat}) \neq 10$ * $n(1-p) ext{ (or } n(1- ext{p-hat})) \neq 10$ * In a confidence interval context, because the true population proportion $p$ is unknown, we use the sample proportion $ext{p-hat}$ ( $\hat{p}$ ) to check this condition.

The Logic of a Confidence Interval

Point Estimate: The sample proportion ( $\hat{p}$ ) acts as a single-number estimate for the population parameter ( $p$ ). It "points" to where the true value might be.
Sampling Distributions: A sampling distribution contains all possible sample proportions of a specific size $n$ taken from the same population. * Center: The mean of all sample proportions is the true population proportion $p$ . * Spread (Standard Deviation): Calculated using the formula $\sigma_{\hat{p}} = \sqrt{\frac{p(1-p)}{n}}$ . * Shape: Normal, provided the large counts condition is met.
The 95% Rule: In a normal distribution, approximately 95% of all possible sample proportions fall within two standard deviations of the center. More accurately, using a Z-table or invNorm, 95% of values fall within $1.96$ standard deviations.
Interval Construction: Since 95% of sample proportions are near the truth, if we take our specific $\hat{p}$ and add/subtract 1.96 standard deviations, we have a 95% chance that our interval will capture the true population proportion $p$ .

Constructing One-Sample Z-Intervals for Proportions

The Problem of the Unknown $p$ : The formula for standard deviation requires the true population proportion $p$ . Since we do not know $p$ , we substitute it with the sample proportion $\hat{p}$ .
Standard Error ( $SE$ ): When we use $\hat{p}$ instead of $p$ in the spread formula, the result is called the standard error ( $SE$ ): * $SE_{\hat{p}} = \sqrt{\frac{\hat{p}(1-\hat{p})}{n}}$
The Critical Value ( $z^$ ): This is the number of standard errors required to reach the desired level of confidence ( $C$ ). Common levels include: 90% Confidence: $z^* = 1.645$ * 95% Confidence: $z^* = 1.96$ * 99% Confidence: $z^* = 2.576$
Calculating $z^*$ : 1. Determine Alpha ( $\alpha$ ): $\alpha = 1 - C$ (where $C$ is the confidence level in decimal form). 2. Split the tails: $\frac{\alpha}{2}$ . 3. Find the Z-score that corresponds to the area of $\frac{\alpha}{2}$ at the bottom of the distribution (using invNorm or a Z-table).
Margin of Error ( $ME$ ): The product of the critical value and the standard error: * $ME = z^* \times \sqrt{\frac{\hat{p}(1-\hat{p})}{n}}$
Final Interval Formula: * $\hat{p} \pm z^* \times \sqrt{\frac{\hat{p}(1-\hat{p})}{n}}$

Four-Step Process Example: Teacher Student Loan Debt

Scenario: A random sample of $780$ teachers in the US found that $82\%$ have college loan debt. Create a $98\%$ confidence interval for the true proportion of all US teachers with debt.

Step 1: State: Identify the procedure and the parameter. * One-sample Z-interval for the proportion of all US teachers with college loan debt.
Step 2: Plan: Check conditions. * Random: Stated in the problem. * Independence: $780$ is less than 10% of all US teachers. * Large Counts: $780 \times 0.82 = 639.6$ (approx. $640$ ) successes; $780 \times 0.18 = 140.4$ (approx. $140$ ) failures. Both are $\geq 10$ .
Step 3: Do: Perform calculations. * $\hat{p} = 0.82$ * $z^$ for 98%: Area at bottom tail is $0.01$ . $z^ = 2.326$ . * $SE = \sqrt{\frac{0.82 \times 0.18}{780}} = 0.01377$ * $ME = 2.326 \times 0.01377 = 0.032$ * Interval: $0.82 \pm 0.032 \rightarrow [0.788, 0.852]$
Step 4: Conclude: Interpret in context. * "I am 98% confident that the true proportion of all teachers in the United States that have student loan debt is somewhere between 78.8% and 85.2%."

Interpreting Confidence Level and Claims

Interpretation of "Confidence Level": A $C\%$ confidence level means that if we were to take many, many samples of the same size and construct intervals for each, about $C\%$ of those intervals would capture the true population proportion. * It is not a probability that the true parameter lies in a specific interval; the parameter is a fixed value. The interval is what varies by sample.
Justifying Claims: * Values inside the interval are considered plausible. * If a claimed value (e.g., 70%) is entirely outside the interval, we have evidence against that claim or evidence that the true value is different from that claim. * In the teacher example, if someone claimed the proportion was over 70%, our interval $[78.8\%, 85.2\%]$ provides evidence for that claim because the entire interval is above 70%.

Determining Sample Size ( $n$ ) for a Desired Margin of Error

To achieve high confidence with a small margin of error ( $ME$ ), researchers need larger sample sizes because larger samples vary less.
Algebraic Process: * Set the formula: $ME = z^* \times \sqrt{\frac{\hat{p}(1-\hat{p})}{n}}$ * The Guess for $\hat{p}$ : If no prior value for $\hat{p}$ is available, use $0.5$ (50%). This is a conservative estimate that yields the largest possible sample size required.
Example Calculation: Find $n$ for 95% confidence and an $ME$ of $0.02$ . 1. $0.02 = 1.96 \times \sqrt{\frac{0.5 \times 0.5}{n}}$ 2. $\frac{0.02}{1.96} = \sqrt{\frac{0.25}{n}}$ 3. $(\frac{0.02}{1.96})^2 = \frac{0.25}{n}$ 4. $n = \frac{0.25}{(\frac{0.02}{1.96})^2}$ 5. $n = 2,401$

Confidence Intervals for the Difference of Two Proportions

Scenario: Estimating the difference between teachers with debt ( $\hat{p}_1$ ) and nurses with debt ( $\hat{p}_2$ ).
Procedure: Two-Sample Z-Interval for $\text{p}_1 - \text{p}_2$ .
Formula: * $(\hat{p}_1 - \hat{p}_2) \pm z^* \times \sqrt{\frac{\hat{p}_1(1-\hat{p}_1)}{n_1} + \frac{\hat{p}_2(1-\hat{p}_2)}{n_2}}$
Interpretation of the Difference: * Entirely Positive Interval: Suggests that the first proportion is significantly larger than the second. * Entirely Negative Interval: Suggests that the first proportion is significantly smaller than the second. * Interval Overlapping Zero: If the interval ranges from negative to positive (e.g., $[-3.2\%, +5.48\%]$ ), zero is a plausible value. This means there is no evidence of a significant difference between the two proportions; they could be the same.

AP Statistics Formula Sheet Guidance

Sampling Distribution Section: Provides the standard deviation (for parameters) and standard error (for intervals) for both single proportions and the difference between proportions.
Generic Formula: The sheet provides a general template: * $\text{Statistic} \pm (\text{Critical Value})(\text{Standard Error})$
Specific Structures: * For one sample: $\hat{p} \pm z^* \times SE_{\hat{p}}$ * For two samples: $(\hat{p}1 - \hat{p}_2) \pm z^* \times SE{\hat{p}_1 - \hat{p}_2}$

Overview of Statistical Inference for Proportions - Definition of Inference: Inference is the process of using sample statistics to make judgments, predictions, or estimates about a larger population parameter. - Context of Proportions: Proportions are primarily derived from categorical data (e.g., the proportion of students who completed homework, or the proportion of males vs. females who did so). - The Two Procedures of Inference: * Confidence Intervals: Used to estimate a population parameter based on a sample statistic. It provides a range of plausible values for the parameter. * Significance Tests: Used to determine the validity of a claim made about a population parameter (i.e., is the claim likely true or false?). - Scope of Application: These procedures can be applied to a single sample from one population or to the difference between two samples from two different populations. ## Necessary Conditions for Inference Before performing any inference procedures for proportions, three conditions must be met to ensure the validity of the results: - Randomness: The sample must be selected randomly to avoid bias. This ensures the sample is representative of the population. - Independence (The 10% Rule): The sample size $n$ must be less than 10% of the total population ( $N$ ). This allows researchers to assume independence between observations when sampling without replacement. - Large Counts (Shape/Normality): The sample size must be large enough so that the sampling distribution of the sample proportion is approximately normal. For proportions, this is satisfied if there are at least 10 successes and 10 failures in the sample: * $np\text{ (or } n\text{p-hat}) \neq 10$ * $n(1-p) \text{ (or } n(1-\text{p-hat})) \neq 10$ * In a confidence interval context, because the true population proportion $p$ is unknown, we use the sample proportion $\text{p-hat}$ ( $\hat{p}$ ) to check this condition. ## The Logic of a Confidence Interval - Point Estimate: The sample proportion ( $\hat{p}$ ) acts as a single-number estimate for the population parameter ( $p$ ). It "points" to where the true value might be. - Sampling Distributions: A sampling distribution contains all possible sample proportions of a specific size $n$ taken from the same population. * Center: The mean of all sample proportions is the true population proportion $p$ . * Spread (Standard Deviation): Calculated using the formula $\sigma{\hat{p}} = \sqrt{\frac{p(1-p)}{n}}$ . * Shape: Normal, provided the large counts condition is met. - The 95% Rule: In a normal distribution, approximately 95% of all possible sample proportions fall within two standard deviations of the center. More accurately, using a Z-table or `invNorm`, 95% of values fall within $1.96$ standard deviations. - Interval Construction: Since 95% of sample proportions are near the truth, if we take our specific $\hat{p}$ and add/subtract 1.96 standard deviations, we have a 95% chance that our interval will capture the true population proportion $p$ . ## Constructing One-Sample Z-Intervals for Proportions - The Problem of the Unknown $p$ : The formula for standard deviation requires the true population proportion $p$ . Since we do not know $p$ , we substitute it with the sample proportion $\hat{p}$ . - Standard Error ( $SE$ ): When we use $\hat{p}$ instead of $p$ in the spread formula, the result is called the standard error ( $SE$ ): * $SE{\hat{p}} = \sqrt{\frac{\hat{p}(1-\hat{p})}{n}}$ - **The Critical Value ( $z^$ ): This is the number of standard errors required to reach the desired level of confidence ( $C$ ). Common levels include: 90% Confidence: $z^ = 1.645$ * 95% Confidence: $z^* = 1.96$ * 99% Confidence: $z^* = 2.576$ - Calculating $z^$ : 1. Determine Alpha ( $\alpha$ ): $\alpha = 1 - C$ (where $C$ is the confidence level in decimal form). 2. Split the tails: $\frac{\alpha}{2}$ . 3. Find the Z-score that corresponds to the area of $\frac{\alpha}{2}$ at the bottom of the distribution (using `invNorm` or a Z-table). - Margin of Error ( $ME$ ): The product of the critical value and the standard error: * $ME = z^ \times \sqrt{\frac{\hat{p}(1-\hat{p})}{n}}$ - Final Interval Formula: * $\hat{p} \pm z^* \times \sqrt{\frac{\hat{p}(1-\hat{p})}{n}}$ ## Four-Step Process Example: Teacher Student Loan Debt Scenario: A random sample of $780$ teachers in the US found that $82\%$ have college loan debt. Create a $98\%$ confidence interval for the true proportion of all US teachers with debt. - Step 1: State: Identify the procedure and the parameter. * One-sample Z-interval for the proportion of all US teachers with college loan debt. - Step 2: Plan: Check conditions. * Random: Stated in the problem. * Independence: $780$ is less than 10% of all US teachers. * Large Counts: $780 \times 0.82 = 639.6$ (approx. $640$ ) successes; $780 \times 0.18 = 140.4$ (approx. $140$ ) failures. Both are $\geq 10$ . - Step 3: Do: Perform calculations. * $\hat{p} = 0.82$ * $z^$ for 98%: Area at bottom tail is $0.01$ . $z^ = 2.326$ . * $SE = \sqrt{\frac{0.82 \times 0.18}{780}} = 0.01377$ * $ME = 2.326 \times 0.01377 = 0.032$ * Interval: $0.82 \pm 0.032 \rightarrow [0.788, 0.852]$ - Step 4: Conclude: Interpret in context. * "I am 98% confident that the true proportion of all teachers in the United States that have student loan debt is somewhere between 78.8% and 85.2%." ## Interpreting Confidence Level and Claims - Interpretation of "Confidence Level": A $C\%$ confidence level means that if we were to take many, many samples of the same size and construct intervals for each, about $C\%$ of those intervals would capture the true population proportion. * It is not a probability that the true parameter lies in a specific interval; the parameter is a fixed value. The interval is what varies by sample. - Justifying Claims: * Values inside the interval are considered plausible. * If a claimed value (e.g., 70%) is entirely outside the interval, we have evidence against that claim or evidence that the true value is different from that claim. * In the teacher example, if someone claimed the proportion was over 70%, our interval $[78.8\%, 85.2\%]$ provides evidence for that claim because the entire interval is above 70%. ## Determining Sample Size ( $n$ ) for a Desired Margin of Error - To achieve high confidence with a small margin of error ( $ME$ ), researchers need larger sample sizes because larger samples vary less. - Algebraic Process: * Set the formula: $ME = z^* \times \sqrt{\frac{\hat{p}(1-\hat{p})}{n}}$ - The Guess for $\hat{p}$ : If no prior value for $\hat{p}$ is available, use $0.5$ (50%). This is a conservative estimate that yields the largest possible sample size required. - Example Calculation: Find $n$ for 95% confidence and an $ME$ of $0.02$ . 1. $0.02 = 1.96 \times \sqrt{\frac{0.5 \times 0.5}{n}}$ 2. $\frac{0.02}{1.96} = \sqrt{\frac{0.25}{n}}$ 3. $(\frac{0.02}{1.96})^2 = \frac{0.25}{n}$ 4. $n = \frac{0.25}{(\frac{0.02}{1.96})^2}$ 5. $n = 2,401$ ## Confidence Intervals for the Difference of Two Proportions - Scenario: Estimating the difference between teachers with debt ( $\hat{p}1$ ) and nurses with debt ( $\hat{p}2$ ). - Procedure: Two-Sample Z-Interval for $\text{p}1 - \text{p}2$ . - Formula: * $(\hat{p}1 - \hat{p}2) \pm z^* \times \sqrt{\frac{\hat{p}1(1-\hat{p}1)}{n1} + \frac{\hat{p}2(1-\hat{p}2)}{n2}}$ - Interpretation of the Difference: * Entirely Positive Interval: Suggests that the first proportion is significantly larger than the second. * Entirely Negative Interval: Suggests that the first proportion is significantly smaller than the second. * Interval Overlapping Zero: If the interval ranges from negative to positive (e.g., $[-3.2\%, +5.48\%]$ ), zero is a plausible value. This means there is no evidence of a significant difference between the two proportions; they could be the same. ## AP Statistics Formula Sheet Guidance - Sampling Distribution Section: Provides the standard deviation (for parameters) and standard error (for intervals) for both single proportions and the difference between proportions. - Generic Formula: The sheet provides a general template: * $\text{Statistic} \pm (\text{Critical Value})(\text{Standard Error})$ - Specific Structures: * For one sample: $\hat{p} \pm z^* \times SE{\hat{p}}$ * For two samples: $$(\hat{p}1 - \hat{p}2) \pm z^ \times SE*{\hat{p