Logistic Regression - Dr. Pam Arroway
- Scheduling for 2009

Course Description :

Six Sigma Black Belts and professionals with an understanding of the fundamentals of least squares regression or analysis of variance are encouraged to attend this course in Logistic Regression. While least squares regression and analysis of variance methods are applicable to data with continuous response variables assumed normally distributed, this course is designed to provide attendees with skills for developing explanatory and predictive models to analyze data with categorical responses. The text for the course is "Introduction to Categorical Data Analysis," by Alan Agresti.

Course content includes:

• Basics of Analysis of Cross-Tabulated Data
• A Review of Issues and Methods in Least Squares Regression Modeling
• Generalized Linear Models
• Statistical Methods for Hypothesis Tests and Interval Estimates with Generalized Linear Models
• Logistic Regression for Analysis of Binomial Response Data
• Logistic Regression for Analysis of Multinomial Response Data
• Logistic Regression for Analysis of Matched Categorical Response Data

Participants are encouraged to bring to class datasets of personal interest with categorical responses and any number of explanatory variables. With the presentation of numerous illustrative examples, there will be ample opportunity to become familiar with computer software routines for the construction and analysis of Logistic Regression Models. There will be a block of time reserved for participants to analyze the data they bring to the class, or other problem datasets to be made available.

Learning Objectives:

• Fundamental concepts of Categorical Data Analysis
• Logistic Regression techniques for relating categorical response variables to explanatory variables
• Knowledge to build, evaluate, and interpret explanatory models of data with categorical responses
• Facility with computing software for the application of logistic regression
• Ability to determine sample size requirements for detecting significant effects on response outcomes
• Insights on relationship patterns in datasets of personal interest