140.753.01
Advanced Methods in Biostatistics III
Location
East Baltimore
Term
3rd Term
Department
Biostatistics
Credit(s)
4
Academic Year
2015 - 2016
Instruction Method
TBD
Tu, Th, 10:30 - 11:50am
Lab Times
Tuesday, 9:00 - 10:20am (01)
Auditors Allowed
Yes, with instructor consent
Available to Undergraduate
No
Grading Restriction
Letter Grade or Pass/Fail
Course Instructor(s)
Contact Name
Frequency Schedule
Every Year
Resources
Prerequisite
140.751-752; Students must also register for 140.754
Introduces generalized linear model (GLM). Foundational topics include: contingency tables, logistic regression for binary and binomial data, models for polytomous data, Poisson log-linear model for count data, and GLM for exponential family. Methods for model fitting, diagnosis, interpretation and inference will be introduced and expanded with techniques for handling overdispersion, quasi-likelihood and conditional likelihood. Also introduces the concept of fixed effects and random effects for modeling clustered data. Related computing techniques including expectation-maximization (EM) and Markov Chain Monte Carlo are covered by an independent computing/data lab series running concurrently with the course.
Learning Objectives
Upon successfully completing this course, students will be able to:
- Use generalized linear model (GLM) to analyze continuous, categorical and count data,
- Know how to construct, fit and interpret different types of GLM in the context of scientific and public health applications
- Understand connections and differences between logistic regression, Poisson log-linear regression and linear regression,
- Conduct statistical inference in these models,
- Diagnose model assumptions,
- Know how to deal with overdispersion in GLM,
- Expand the model and inference tools with quasi-likelihood and conditional likelihood,
- Extend linear model to account for clustering using random effects,
- Apply theoretical concepts to scientific data using R software,
- Improve computational and analytic skills through analysis of simulated and real data sets,