Class 8B - ph 45

Page 1: Class&Introduction

PH 45 Class 8B

Page 2: Agenda

Overview of Meeting Topics

1. Extra Credit Opportunity

2. Group Assessment

3. Review of Bivariate Analyses

4. Regression Analysis

5. Updates and Reminders

Page 3: Extra Credit Opportunity

Introduction to Extra Credit

Page 4: Details of Extra Credit Opportunity

Conditions for Extra Credit

• Correction: PIDs are collected.

• Participation Requirement: If 80% participation is achieved, every student will receive 2 points of extra credit.

• Encourage peers to complete the survey.

Page 5: Importance of Student Voices

Purpose of the Survey

• Goal: Understand financial insecurity among UCSD students.

• Method: Brief anonymous survey to gather data on financial challenges.

• Outcome: Findings will be presented to UCSD leadership to guide interventions.

• Access: Survey available on computer, phone, or tablet with QR code/link provided.

• Organized by: The 2025 BSPH Honors Practicum cohort.

Page 6: Group Assessment

Upcoming Group Assessment

Page 7: Peer Evaluation & Assignment Grading

Group Assessment Guidelines

• Feedback: Anonymous feedback from each group member on teammates’ contributions.

• Impact on Grades: Feedback will influence final grade on the assignment.

• Accountability: Active contributors earn full participation credit; non-contributors may face grade deductions.

• Focus: Emphasis on fairness, teamwork, communication, and support.

Page 8: Bivariate Analyses Review

Overview of Bivariate Analyses

Page 13: Regression Analysis

Overview

Page 14: Introduction to Regression Analysis

Understanding Regression

• Purpose: Models relationships between variables and predicts outcomes.

• Usage in Public Health: Analyzes health trends, risk factors, and policy development.

• Significance: Critical tool in epidemiology for understanding cause-and-effect.

Page 15: Types of Regression

Types and Applications

• Linear Regression: Predicts a continuous outcome.

• Logistic Regression: Predicts a categorical outcome.

• Multiple Regression: Uses multiple predictors for both linear and logistic regression.

Page 16: Variables in Regression

Independent vs. Dependent Variables

• Independent Variable: Factors influencing outcomes (e.g., smoking status).

• Dependent Variable: Results predicted or explained (e.g., lung cancer diagnosis).

• Function of Regression: Analyzes influence of variables while adjusting for multiple factors.

Page 17: Linear Regression Definition

Characteristics of Linear Regression

• Definition: Statistical method modeling relationship between independent and dependent variables.

• Use Case: Continuous outcome examples (e.g., blood pressure).

• Common Applications: Assess obesity rates, hospital readmission predictions, impact of air pollution.

Page 18: Logistic Regression Definition

Characteristics of Logistic Regression

• Definition: Method for categorical outcomes prediction (e.g., disease presence).

• Use Case: Predict probabilities of health outcomes.

• Common Applications: Assess smoking cessation likelihood, heart disease risk factors, vaccine uptake influence.

Page 21: Assumptions of Linear Regression

Key Assumptions

1. Linearity: Relationship must be linear.

2. Independence: Observations should not correlate.

3. Homoscedasticity: Residuals’ variance should be constant.

4. Normality: Residuals should follow a normal distribution.

5. Robustness: Minor violations are acceptable; major require adjustments.

Page 22: Assumptions of Logistic Regression

Key Assumptions

1. Linearity in Log-Odds: Independent variables relate linearly to log-odds.

2. Independence of Observations: Data points must be independent.

3. Large Sample Size: Enhances model estimates.

4. Robustness: Less sensitive to outliers than linear regression.

Page 27: Directionality in Regression Models

Regression vs. Correlation

• Association vs. Causation: Correlation shows association but not direction.

• Directional Relationships: Regression establishes influence (e.g., X influences Y).

• Key Takeaway: Correlation does not equate to causation.

Page 28: Directionality in Regression vs. Bivariate Analysis

Differences in Assumptions

• Regression: Assumes directional relationships.

• Bivariate Analysis: Does not assume directionality.

• Importance of Correct Methods: Regression used for examining impact; bivariate for exploring relationships without inferring causation.

Page 29: Bivariate vs. Multivariate Analysis

Concepts and Applications

• Bivariate Analysis: Examines relationship between one independent and dependent variable.

• Multivariate Analysis: Involves several predictors to assess collective effects on a dependent variable.

• Key Differences: Multivariate controls confounding variables for deeper insights.

Page 30: Benefits of Multivariate Analysis

Importance in Research

• Confounding Variables: Adjusts for third variables potentially influencing relationships.

• Modeling Accuracy: Improves clarity on influences affecting outcomes.

• Public Health Applications: Evaluates risk factors for diseases and vaccination hesitancy, among others.

Page 31: Interpretation of Bivariate and Multivariate Analyses

Key Insights

• Bivariate Interpretation: Direct effect of one predictor; e.g., smoking’s odds ratio in predicting lung cancer.

• Multivariate Interpretation: Adjusted estimates with confounders considered; e.g., smoking’s odds after factoring in age and pollution.

Page 32: Key Takeaways on Analyses Types

Final Thoughts

• Bivariate Analysis: Useful but may mislead if confounders aren't noted.

• Multivariate Analysis: Offers robustness and adjusted estimates reflecting causal links.

Page 54: Updates and Reminders

Recap of Key Information

Page 55: Detailed Updates and Reminders

Important Dates and Information

• Office Hours: Schedule in advance due to limited availability.

• Draft Submission Deadlines:

  • Table 2: Rough Draft due March 3rd by 5:00 pm (3 points).

  • Table 3: Rough Draft due March 7th by 5:00 pm (4 points).