Chapter 8: Confidence Intervals – Business Statistics Study Notes

Confidence Intervals

Introduction to Confidence Intervals

Confidence intervals are a fundamental concept in business statistics, providing a range of values within which a population parameter is expected to lie, based on sample data. This chapter focuses on the construction, interpretation, and application of confidence intervals in business contexts.

• Point Estimate: A single value (such as the sample mean or proportion) used to estimate a population parameter.

• Confidence Interval (CI): An interval estimate that gives a range of plausible values for the population parameter, reflecting the uncertainty inherent in sampling.

• Margin of Error (ME): The amount added and subtracted from the point estimate to create the confidence interval; it quantifies the uncertainty due to sampling variability.

Example: The Bureau of Labour Statistics (BLS) reports monthly unemployment rates based on sample surveys. These rates are point estimates and are subject to sampling error. Reporting a confidence interval alongside the point estimate provides a more complete picture of uncertainty.

Difference Between Point Estimate and Confidence Interval

Definitions and Importance

A point estimate is a single value that best describes the population parameter of interest, such as the sample mean ($\bar{x}$) or sample proportion ($\hat{p}$). However, point estimates do not convey information about their accuracy or reliability.

• Point Estimate: Easy to calculate, but does not indicate how close it is to the true population value.

• Confidence Interval: Provides a range of values, indicating the degree of uncertainty and the likelihood that the interval contains the true population parameter.

Example: If a sample mean of hospital stay is 5.11 days, the point estimate is 5.11. A confidence interval might be (4.64, 5.58), indicating the true mean is likely within this range.

Calculating Confidence Intervals for the Mean

When Population Standard Deviation ($\sigma$) is Known

When the population standard deviation is known, the confidence interval for the mean is calculated using the normal (z) distribution.

• Formula:

$\text{CI} = \bar{x} \pm z_{\alpha/2} \left( \frac{\sigma}{\sqrt{n}} \right)$

• $\bar{x}$ = sample mean

• $z_{\alpha/2}$ = critical z-score for desired confidence level

• $\sigma$ = population standard deviation

• $n$ = sample size

Example: For a 95% confidence level, $z_{\alpha/2} = 1.96$.

When Population Standard Deviation ($\sigma$) is Unknown

If $\sigma$ is unknown, use the sample standard deviation ($s$) and the t-distribution.

• Formula for Sample Standard Deviation:

$s = \sqrt{\frac{\sum (x_i - \bar{x})^2}{n-1}}$

• $x_i$ = each data point

• $\bar{x}$ = sample mean

• $n$ = sample size

• Confidence Interval Formula (t-distribution):

$\text{CI} = \bar{x} \pm t_{\alpha/2, df} \left( \frac{s}{\sqrt{n}} \right)$

• $t_{\alpha/2, df}$ = critical t-score for desired confidence level and degrees of freedom ($df = n-1$)

• $s$ = sample standard deviation

Properties of the t-distribution:

• Bell-shaped and symmetrical

• Flatter and wider than the normal distribution for small sample sizes

• As sample size increases, t-distribution approaches the normal distribution

Example: For $n = 15$, $df = 14$, and a 95% confidence level, $t_{0.025, 14} \approx 2.145$.

Calculating Interval Endpoints

• Upper Confidence Limit (UCL): $\bar{x} + t_{\alpha/2, df} \left( \frac{s}{\sqrt{n}} \right)$

• Lower Confidence Limit (LCL): $\bar{x} - t_{\alpha/2, df} \left( \frac{s}{\sqrt{n}} \right)$

Example: If $\bar{x} = 5.11$, $s = 0.85$, $n = 15$, $t_{0.025, 14} = 2.145$:

$\text{Margin of Error} = 2.145 \times \frac{0.85}{\sqrt{15}} = 0.471$

$\text{CI} = (5.11 - 0.471, 5.11 + 0.471) = (4.64, 5.58)$

Calculating Confidence Intervals for Proportions

Confidence Interval for a Proportion

Proportion data often follow the binomial distribution. For large samples, the normal approximation is used if $np \geq 5$ and $n(1-p) \geq 5$.

• Sample Proportion: $\hat{p} = \frac{x}{n}$

• Standard Error of Proportion:

$SE_{\hat{p}} = \sqrt{\frac{\hat{p}(1-\hat{p})}{n}}$

• $x$ = number of successes

• $n$ = sample size

• Confidence Interval Formula:

$\text{CI} = \hat{p} \pm z_{\alpha/2} \times SE_{\hat{p}}$

• $z_{\alpha/2}$ = critical z-score

Example: In a sample of 100 U.S. citizens, 22 have blue eyes. $\hat{p} = 0.22$. For a 98% confidence level, $z_{\alpha/2} = 2.33$.

$SE_{\hat{p}} = \sqrt{\frac{0.22 \times 0.78}{100}} = 0.0414$

$\text{CI} = 0.22 \pm 2.33 \times 0.0414 = (0.123, 0.317)$

Margin of Error and Sample Size

Margin of Error (ME)

The margin of error quantifies the uncertainty in the estimate and is used to determine the width of the confidence interval.

• For Means: $ME = t_{\alpha/2, df} \left( \frac{s}{\sqrt{n}} \right)$ or $ME = z_{\alpha/2} \left( \frac{\sigma}{\sqrt{n}} \right)$

• For Proportions: $ME = z_{\alpha/2} \times SE_{\hat{p}}$

Application: Determining the required sample size for a desired margin of error is a key aspect of survey design.

Interpreting Confidence Intervals

Statistical Meaning

A confidence interval does not guarantee that the true parameter is within the interval for a given sample. Instead, it means that, over many repeated samples, a certain percentage (e.g., 95%) of the intervals constructed will contain the true parameter.

• Confidence Level: The probability that the interval contains the true parameter in repeated sampling (e.g., 95%).

• Reporting: Always report both the point estimate and the confidence interval to convey uncertainty.

Example: "We are 95% confident that the average hospital stay is between 4.64 and 5.58 days."

Practical Applications in Business Statistics

Use in Surveys and Analytics

Confidence intervals are widely used in business analytics, especially in survey research and reporting economic indicators. They provide a statistical basis for decision-making under uncertainty.

• Surveys: Organizations like Pew Research Center and Google report survey results with confidence intervals and margin of error to indicate reliability.

• Economic Indicators: Agencies such as the BLS use sample surveys to estimate unemployment rates, reporting confidence intervals to reflect sampling variability.

Table: Comparison of Critical Values for t and z Distributions

Confidence Level (%)	t (df=10)	t (df=20)	t (df=30)	z
80	1.372	1.325	1.310	1.28
90	1.812	1.725	1.697	1.645
95	2.228	2.086	2.042	1.96
99	3.169	2.845	2.750	2.575

Additional info: As sample size increases, the t-distribution approaches the normal (z) distribution, and the critical values become nearly identical.

Summary

Confidence intervals are essential tools in business statistics for quantifying uncertainty in estimates of population parameters. They are constructed using sample data, with formulas depending on whether the population standard deviation is known or unknown, and are interpreted in the context of repeated sampling. Reporting both point estimates and confidence intervals provides a more complete and honest representation of statistical findings in business and economics.