GMT20241028-215824_Recording_avo_640x360

Attendance & Course Updates

  • Several students called out during attendance.

  • Reminder: Homework due this Sunday at midnight from chapters 6 and 7 (only 3 questions).

  • Quiz 5 scheduled for next Monday covering chapter 7.

Recap of Previous Topics

Estimation and Sampling

  • Last topic covered was estimation from samples to populations.

  • Working with sample statistics to infer population parameters such as mean and variance.

  • Bias and Unbiased Estimators: The difference between expected value of estimators and true population value determines bias.

Variance and Standard Error

  • Smaller variance in estimators indicates a better estimator.

  • Mean Square Error (MSE) combines variance and bias.

Central Limit Theorem (CLT)

  • Any mean derived from a random sample of a population follows a normal distribution irrespective of the original population distribution.

  • Discussed methods to find estimators, including Moment Estimation and Maximum Likelihood Estimation (MLE).

Maximum Likelihood Estimation (MLE)

  • MLE estimates population parameters by maximizing the likelihood function derived from the sample data.

  • Likelihood function combines probabilities of observed values in the sample.

  • The estimator is the parameter value that maximizes the likelihood given the observed data.

Example of MLE:

  • Example with the weights of American female college students derived a mean from a sample and the maximum likelihood estimate was calculated.

  • For problems with known standard deviations, the MLE can be computed to provide estimates for unknown parameters.

Hypothesis Testing Introduction

  • Importance of hypothesis testing in statistics for confirming or rejecting claims.

  • Null Hypothesis (H0): Statement of no effect or no difference.

  • Alternative Hypothesis (H1): The claim that contradicts the null hypothesis.

    • Example: Testing if the average GPA at GSU differs from a specific value (2.0).

Outcomes of Hypothesis Testing

  • Four possible outcomes:

    • True H0 & Fail to Reject: Correct outcome.

    • True H0 & Reject: Type I Error (False Positive).

    • False H0 & Fail to Reject: Type II Error (False Negative).

    • False H0 & Reject: Correct outcome.

Rejection Criteria

  • Decision making based on sample means and predetermined significance levels.

  • Use of critical values from statistical distributions to determine rejection or acceptance of H0.

Steps in Hypothesis Testing

  1. Identify the parameter of interest (mean, proportion, etc.).

  2. State H0 and H1.

  3. Decide on the significance level (usually 0.05).

  4. Determine appropriate test statistics (e.g., z-test).

  5. Define rejection region based on significance level.

  6. Collect sample data and make decisions based on test statistic and rejection region.

Example Test Statistics

  • Worked through an example calculating sample mean and determining if H0 could be rejected:

N=5, Sample mean = 51.3, Claimed population mean = 50, Reject H0 if Z value exceeds certain critical values.

P-Value Methodology

  • Introduced the concept of the p-value as an alternative to traditional rejection region methodology.

  • P-value directly indicates whether the null hypothesis can be rejected based on the significance level; if p-value < alpha, reject H0.

  • Discussed the differences in p-values for one-tailed and two-tailed tests.

Conclusion

  • Next session will involve detailed practical applications of p-values and hypothesis testing in real-world examples.

  • Important to understand the relationship between samples, parameters, and decision-making in statistical contexts.