BIOE 850 - Categorical Data Analysis

Cr Hrs: 2 (32-0-0)
This course begins by an introduction and review of most common discrete random variables and their probability distributions, followed by a brief discussion of ‘parameter estimation’ as a general concept in Theoretical Statistics. Then, we introduce the concept of inferential statistics by discussing one sample confidence interval and hypothesis testing for one- and two-sample designs, which includes the definition of and testing for statistical independence through the most commonly used chi-squarebased tests for 2x2, Rx2, 2xC, and RxC contingency tables and sets of (stratified) contingency tables. Then, the generalized linear model is introduced as the backbone for model building that focuses on the estimation of effects of one or more predictors on a binary response variable or on a count variable, including model inference and model diagnostics checking. Specific topics for the modeling of categorical data include logistic regression for dichotomous and polytomous response, conditional logistic regression, generalized estimating equations, and generalized linear mixed modeling for models with random effects. In addition, the course will explore log-linear modeling for count data. The relation of the various approaches and procedures using SAS will be demonstrated. The course focuses on application of the above approaches to observational and clinical trial designs.

Grade Mode: Standard

Mode of Delivery (Online or Hybrid): Hybrid
Instructional Method: Lecture
Prerequisites: BIOE 812 and BIOE 821 or approval by instructor

Term offered: Fall

Fall - Instructor of Record: Mehmet Kocak