Tufts OpenCourseware
Search
HOME|SCHOOLS|COURSES|ABOUT OCW| TUFTS OER| HELP|FEEDBACK|CONTACT|
Tufts University
Please send us your feedback
» OCW Home » Epidemiology and Biostatistics » Lectures 
Lecture 12: Regression II: The Realities of Model Building
Author: P. Newby, Sc.D.,M.P.H.
Get PDF Version of Slides | About PDF Format
Lecture 12 - Introduction Slide Lecture 12 - Introduction Slide
Objectives Objectives
Outline for lecture Outline for lecture
Our example dataset Our example dataset
Application of Concepts Application of Concepts
First things first First things first
Phase 1: Pre-analysis Phase 1: Pre-analysis
Is your dataset “clean”? Is your dataset “clean”?
Data cleaning process Data cleaning process
Employ study restrictions Employ study restrictions
Phase 2: Analyzing a dataset Phase 2: Analyzing a dataset
Step 1: Conduct descriptive statistics Step 1: Conduct descriptive statistics
Example: Table 1 – Sample characteristics Example: Table 1 – Sample characteristics
Review questions for Table 1 Review questions for Table 1
Step 2: Conduct additional tests to describe the data Step 2: Conduct additional tests to describe the data
Example:Table 2 – Sample characteristics by exposure group (dietary pattern) Example:Table 2 – Sample characteristics by exposure group (dietary pattern)
Review questions for Table 2 Review questions for Table 2
Step 3: Building a regression model Step 3: Building a regression model
Building a model involves many steps Building a model involves many steps
Step 3.1: Build the core regression model Step 3.1: Build the core regression model
Univariate regression results Univariate regression results
Step 3.2 Testing for confounding Step 3.2 Testing for confounding
Review questions - confounding Review questions - confounding
Step 3.3: Begin building the multivariate model Step 3.3: Begin building the multivariate model
Example: core model plus age Example: core model plus age
Continue building the model (1) Continue building the model (1)
Continue building the model (2) Continue building the model (2)
Example: core model versus multivariate adjusted model Example: core model versus multivariate adjusted model
Step 3.4: Testing for interaction Step 3.4: Testing for interaction
Example – interaction test Example – interaction test
Step 3.5:Refine and select the final model Step 3.5:Refine and select the final model
Step 3.6:Presenting the final model Step 3.6:Presenting the final model
Example:Table 4 - Regression models Example:Table 4 - Regression models
Final models in our example Final models in our example
Review questions – interpreting regression coefficients Review questions – interpreting regression coefficients
Conclusions Conclusions
People who viewed this content, also viewed...
Legal Notices|Privacy|Tufts University © 2005-2008