SAS: Apply & Evaluate Poisson & Negative Binomial Models

Overview

This course equips learners with the knowledge and practical skills to analyze, construct, and evaluate statistical models for count data using SAS. Beginning with Poisson regression, learners will identify appropriate datasets, assess distributional assumptions, and build models using PROC GENMOD with the log link function. They will then examine model diagnostics to detect issues such as overdispersion and refine models for better accuracy. Building on these foundations, learners will differentiate between Poisson and negative binomial regression approaches, interpret the role of the dispersion parameter, and compare models using statistical criteria like AIC and deviance. Real-world examples and guided SAS implementations will allow learners to apply these techniques effectively, justify model selection decisions, and optimize predictive performance for diverse count data scenarios. By the end of the course, participants will be able to select, implement, and critique regression models that best fit the characteristics of their datasets, enhancing their analytical capabilities in statistical modeling with SAS.

Syllabus

Poisson Regression Fundamentals

This module introduces learners to the principles and application of Poisson regression using SAS. It covers the fundamentals of modeling count data, exploring datasets for suitability, fitting models using PROC GENMOD, and interpreting results. Learners will progress from understanding the problem context and dataset structure to building and refining Poisson regression models, diagnosing potential issues like overdispersion, and improving model accuracy through variable selection and statistical analysis techniques.

Negative Binomial Regression Techniques

This module builds on the foundations of Poisson regression by introducing the negative binomial model for count data exhibiting overdispersion. Learners will explore when and why to choose the negative binomial approach, understand the role of the dispersion parameter, and evaluate model outputs using statistical diagnostics and information criteria. Through practical SAS implementations, learners will gain the skills to refine models, address influential observations, and compare performance against Poisson regression to select the most appropriate modeling strategy.