stats exam 3
Essentials of Business Statistics: Communicating with Numbers
7.1 Sampling
Statistical Inference: A major portion of statistics involves estimating population parameters and testing hypotheses about them.
- Population: All items of interest in a statistical problem; if the entire population were accessible, parameters would be known, removing the need for inference.
- Sample: A subset of the population.
- Sample statistics are used to infer about unknown population parameters.
- Quality of inference depends on sample credibility; the sample must represent the population.Bias: The tendency of a sample statistic to systematically overestimate or underestimate a population parameter.
- Causes of bias:
- Samples not representative of the population.
- Selection bias: Systematic underrepresentation of certain groups in the sample.
- Nonresponse bias: Systematic differences between respondents and nonrespondents to a survey.
- Social-desirability bias: Respondents providing answers they believe are socially acceptable instead of their true preferences.Good Sample Characteristics:
- A “good” sample must be representative of the population.
- Simple Random Sample: A sample of n observations where each has an equal probability of selection.
- Most statistical methods assume a simple random sample.
- Stratified Random Sample: Population divided into mutually exclusive and collectively exhaustive groups (strata), with simple random samples drawn from each stratum based on proportional sizes.Advantages of Stratified Random Sampling:
- Representation of population subdivisions of interest.
- Greater precision in parameter estimates compared to simple random sampling.
7.2 The Sampling Distribution of the Sample Mean
Population Parameter: A constant value describing a characteristic of the population; unknown parameters include population mean and proportion.
Sample Statistic: A random variable whose value depends on sample choice—used to estimate population parameters.
- Estimator/Point Estimator: A statistic used to estimate a population parameter; for example, the sample mean (ar{x}) estimates the population mean (μ).
- An estimate is a specific value derived from the estimator.Expected Value and Variance:
- For a sample mean based on a random sample of n observations:
- Expected Value (mean of sample means) equals the population mean.
- Unbiased Estimator: Expected value of an estimator equals the population parameter.Standard Error of the Sample Mean (SEM):
- Variance in sample means ensures that high and low values cancel each other out. The standard error quantifies the dispersion of sample means around the population mean.Normal Distribution Requirement:
- Central Limit Theorem (CLT): States that for a given population with an expected value (μ) and standard deviation (σ), the sampling distribution of the sample mean (ar{x}) will approach normal distribution as sample size (n) increases; particularly valid if n ≥ 30.Sampling Distribution Characteristics:
- When the population has a different distribution (not normal), the sample size must be sufficiently large for the sampling distribution to assume normality.
7.3 The Sampling Distribution of the Sample Proportion
Population Proportion (p): The proportion of successes within the population, typically assessed using binomial distribution.
- Sample Proportion (p-hat) (ar{p}): serves as the point estimator for p. The expected distribution approximates normality when conditions of np ≥ 5 and n(1-p) ≥ 5 hold.
- Central Limit Theorem for Sample Proportion:
- As n increases, the sampling distribution of ar{p} approaches normal distribution, particularly as the population proportion departs from 0.50, necessitating larger sample sizes for valid approximations.
7.4 Statistical Quality Control
Control Charts: Tools for monitoring statistical process control:
1. Plot sample estimates and assess if within predetermined guidelines.
2. Reference the centerline of expected value during process control assessment.
3. Establish upper (UCL) and lower (LCL) control limits to signify excessive deviation from expected values:
- UCL = Expected Value + 3 × Standard Error
- LCL = Expected Value - 3 × Standard Error
- Validity of control charts is contingent on the normality of the sampling distribution of the estimator.
8.1 Confidence Interval for the Population Mean When σ Is Known
Population and Sample Definitions:
- Population refers to all items of interest while a sample represents a subset. Inferential statistics revolve around two methodologies: estimation and hypothesis testing.Point Estimator: A statistic used to estimate population parameters; e.g., sample mean (μ-hat) estimates population mean (μ) and sample proportion (p-hat) estimates population proportion (p).
Confidence Interval:
- Rather than point estimates, confidence intervals provide a range of values for population parameters with a defined level of confidence, contingent upon the assumption that the sampling distribution is approximately normal. It can be defined as:
- Point Estimate ± Margin of Error
- The margin accounts for the standard error of the estimate and the confidence level desired.Calculating Confidence Intervals:
- Values are derived from critical values associated with the standard normal random variable Z. For example, for a 95% confidence interval, critical Z-values are approximately ±1.96.Interval Statements:
- Isolate interval according to the sample's nature, with a focus on interchanging sample means drawn repeatedly landing within the defined boundaries at the designated confidence level (e.g., 95%).
8.2 Confidence Interval for the Population Mean When σ Is Unknown
When σ is unknown, the Student's t distribution is utilized:
- T Distribution Characteristics:
- Bell-shaped and symmetric but with broader tails than the z-distribution; the degrees of freedom (df) determine tail broadness.
- As df increase, the t-distribution approximates the z-distribution more closely.Calculating Sample Size:
- Use T.INV for cumulative probability; for instance, T.INV(0.95, df).
- Example Given: MPG follows normal distribution.
- Determine confidence intervals based on a defined standard deviation and sample mean to establish confidence interval parameters, ensuring the construction is sound within specified confidence levels (e.g., 90%).
8.3 Confidence Interval for the Population Proportion
Sample Proportion (ar{p}): Used as the estimator for population proportion p. A 100% confidence interval is only valid if approximately follows a normal distribution based on the conditions provided (np ≥ 5 and n(1-p) ≥ 5).
8.4 Selecting the Required Sample Size
Confidence intervals defined as point estimates ± margin of error define precision:
- This precision can be enhanced with larger sample sizes, although increased sizes also involve heightened costs.
- An ideal preliminary step involves determining minimal sample size for desired margin of error using prior data.Example Given: If limiting margin of error (E) to a defined value (e.g., 2 MPG), calculate minimum sample size needed for achieving stated precision requirements.
9.1 Introduction to Hypothesis Testing
Hypothesis Testing: Involves proposing a statement regarding population parameters that can be substantiated or refuted through evidence.
Null Hypothesis (H0): Default state/assumption of no effect or no difference; alternative hypothesis (HA) opposes the null.
- Null and alternative hypotheses must be clearly outlined as one ultimately influences the conclusions drawn from testing.Types of Tests:
- One-tailed Tests: Only rejecting the null on one side of the hypothesized value of the population parameter.
- Two-tailed Tests: Rejections are possible on either side, when the alternative is ≠.Error Types:
- Type I Error: Rejecting null when true.
- Type II Error: Not rejecting null when it is false.
9.2 Hypothesis Testing for Population Means
Procedure:
1. Define null and alternative hypotheses.
2. Identify significance level.
3. Compute test statistic and p-value from the sample data.
4. Compare p-value to significance level and draw conclusions based on the findings.P-value Explanation: Determines the probability of observing test results given that the null hypothesis is true.
Examples Given: Outline steps in hypothesis tests, including specification of claims about population parameters with appropriate calculations and decisions based on numerical evidence.