Regression For Correlated Data.pdf

Course Title: Regression Methods for Correlated Data
Course Code: STAT/BIOST 571
Semester: Winter 2019
Lectures: Mon/Wed/Fri, 1:30-2:20 PM
- Mon & Fri: South Campus Center (SOCC-301)
- Wed: Health Sciences (K-069)
Credits: 3

Instructor:
- Adam A. Szpiro ([email protected])
- Office Hours: Wed, 2:30-3:30 PM or by appointment (HSB F-673)

Teaching Assistants:
- Xiaowen Tian ([email protected]); Office Hours: Thur, 2:00-3:00 PM (Health Sciences Library)
- Hongjian Shi ([email protected]); Office Hours: Tue, 3:00-4:00 PM (CMU B023)

Course Description:
This course extends regression methods for independent outcomes (covered in STAT/BIOST 570) to handle correlated data. It focuses on application and interpretation of methods in practical circumstances, requiring statistical insight, mathematical and computational skills, and effective communication.

Prerequisites:
- STAT/BIOST 570
- STAT 512/513 or equivalent
- Familiarity with R

Learning Objectives:
- Explain why conventional regression models are inappropriate for correlated data.
- Describe key differences between generalized estimating equations and likelihood approaches.
- Suggest appropriate modeling approaches, evaluate fit, and interpret regression coefficients.
- Outline implementation strategies for various approaches, including Bayesian methods.

Textbooks (on reserve at Health Sciences Library):
- Diggle, Heagerty, Liang, and Zeger (2013). Analysis of Longitudinal Data (second edition)
- Wakefield (2013). Bayesian and Frequentist Regression Methods

Software: R (http://cran.r-project.org/)

Assessment:
- Homework (20%): Weekly assignments, submitted online and due at the beginning of lecture the following week.
- Class Participation (15%): Mandatory; expected to participate about once per week.
- Midterm Exam (30%): One hour, in-class, closed book (Feb 11, SOCC 301)
- Final Exam (35%): Two hours, in-class, closed book (TBD)

Homework & Exams:
- Word processed, using LaTeX recommended.
- Include annotated R code for all problems.
- Write in full sentences and paragraphs; raw code/output not accepted.
- In-class exams: Neat and legible responses required.

Course Discussion Board: Primary forum for electronic questions about course material or homework.

Access & Accommodations: Contact Disability Resources for Students (DRS) for accommodations.