Biostatistics 411 Analysis of Correlated Data

Winter 2013 - Robert Weiss

News & Announcements:

• Data Analysis Project 3 is now posted.
• Example of using SAS to draw a smooth loess curve through a scatterplot.
• The article mentioned in class on 1/30/2013: Wainer, Howard (1997). Improving Tabular Displays, With NAEP Tables as Examples and Inspirations. Journal of Educational and Behavioral Statistics, 22, 1-30.
• Updating for 2013 is complete.

Class Schedule, Email Addresses, and Office Hours:

Syllabus and Textbook

• 2013 Syllabus
• 2013 Course Information - including schedule, office hours, text information, grading policy, SAS and computer information, handout policy.
• Our textbook's web page Fitzmaurice, Laird and Ware (2011). Applied Longitudinal Analysis, 2nd edition.
• The first edition web page Fitzmaurice, Laird and Ware (2004). Applied Longitudinal Analysis.
• HLM Links - Has links to several helpful resources on Hierarchical Linear Models.
• Useful SAS Links - Has links to several helpful sites.

Readings Our text is Applied Longitudinal Analysis (ALA), 2nd edition by Fitzmaurice, Laird and Ware. Readings are updated for 2013.

• Immediately: Appendices A & B in ALA. If you are not perfectly comfortable with vector and matrix addition and multiplication, please read appendix A immediately! Similarly, read appendix B for a review of the properties of expectation and variance.
• Week 1: Chapters 1 and 2 in ALA.
• Week 2: For lab, look through Singer's paper, link below.
• Week 2: Chapters 3, 4 and 5 in ALA. Chapter 4 may be tough for many. Please do your best, we'll talk about it in class, this chapter is not a make or break part of the material.
• Week 3: Chapters 6 and 7 in ALA.
• Week 3: Overview and Getting Started Chapters in the Proc Mixed Documentation. You may skim parts of the overview that are difficult to read.
• Week 4: Chapters 8 and 11 in ALA. Chapter 9 for additional useful information.
• Week 5: Chapters 12, 13 and 14 in ALA.
• Week 5: Overview and Getting Started Chapters in Glimmix Documentation. You may skim parts of the Overview that are difficult to read.
• Week 6: Chapter 16 in ALA.
• Week 6: Overview and Getting Started Chapters in Genmod Documentation. You may skim parts that are difficult to read.
• Week 8: Chapters 21 and 22 in ALA.
• Week 8: Chapter 9 "Random Effects Models" in Weiss (2005) Modeling Longitudinal Data provides an additional perspective on hierarchical models. Available online through the UCLA library at Scroll down and click onto Springer eBooks (2005-12). Search for "Modeling Longitudinal Data" and click on the first book that appears. Scroll down to chapter 9.
• Week 10: Chapter 13 in Weiss (2005) Modeling Longitudinal Data teaches bivariate longitudinal data. Available online through the UCLA library at Scroll down and click onto Springer eBooks (2005-12). Search for "Modeling Longitudinal Data" and click on the first book that appears. Scroll down to chapter 13.
• Comparable readings for the first edition of ALA.

Handouts

• Singer's paper on using SAS Prox Mixed

Homeworks

• Homework 1. Due: Feb 4.
• Small mice data
• Dental data
• Homework 2. Due: Feb 20. [Due date changed to 20th.]
• Bolus Count Data
• Morbidity Data
• How to plot means from a fitted model. One way to plot the results for problem 2. This plot works well as a supplemental plot to all parts of the problem.
• Homework 3. Due: Feb 25.
• Morbidity Data

Data Analysis Projects (DAPs)

• Data Analysis Project 1 . Analysis of Girls Growth Data. Due: Monday, March 4 in class. (20130219 09:30)
• Girls' heights for Data Analysis Project 1 (DAP1)
• 2013 Data Analysis Project 2. Due: Monday, March 11 in class.
• BSI Data for Data Analysis Project 2 (DAP2)
• Data Analysis Project 3. Hardcopy (no emails!) Due: Wednesday, March 20, 3pm in Weiss's mailbox in the Biostat office CHS 51-254.
• Kenya Morbidity Data for Data Analysis Project 3 (DAP3)

Lecture Notes

• Lecture 1. Introduction to Longitudinal Data.
  • Lecture Notes 1: Introduction.
  • Figures for Lecture Notes 1
• Lecture 2 notes. Key Concepts for Longitudinal Data
  • Notes 2 (pdf)
  • SAS code and results (word)
  • SAS code and results with comments (word) (20130109 11:19)
• Lecture 3 notes: Notation, Statistical Concepts, Theory.
  • Notes 3 (pdf) Notation and Statistical Concepts
• Lecture 4 notes. Models for the Mean over Time.
  • Notes 4 (pdf)
  • SAS code and Output
  • Example comparing linear time trend to unstructured time trend (word).
• Lecture 5. Patterned Covariance Models.
  • Notes 5 (pdf)
  • SAS code, output and discussion. (word)
  • Fitting covariance models excel file example.
• Lecture 6. Random Effects Models.
  • Notes 6 (pdf)
  • SAS code, output and discussion. (word)
  • Figures.
• Lecture 7. Generalized Linear Models.
  • Notes 7 (pdf) Generalized Linear Models.
• Lecture 8. Updated for 2013.
  • Notes 8 (pdf) Glimmix
  • Example (word) output
  • Example (excel) output
  • (word) examples useful for homework 1. output
• Lecture 9. Updated for 2013.
  • Notes 9 (pdf) GLMM for Count Data
  • Glimmix output (word) Count Data
• Lecture 10. Generalized Estimating Equations. Updated for 2013.
  • Notes 10 (pdf) Generalized Estimating Equations.
  • Genmod output (word) Count Data
• Lecture 11. Introduction to hierarchical models.
  • Notes 11 (pdf)
  • Examples (word)
• Lecture 12.
  • Notes 12 (pdf) Continuous hierarchical longitudinal data, practical issues.
  • Figures and SAS output(word) Lecture 12.
  • Paper Bursch et al (pdf) For Lecture 12.
• Lecture 13. Hierarchical longitudinal models with glimmix.
  • Notes 13 (pdf)
  • 13 SAS output (word)
• Lecture 14. Hierarchical longitudinal, and a clinical example.
  • Notes 14 (pdf) TVSFP Hierarchical longitudinal data.
  • Notes 14 (pdf) clinical example.
  • Output (word)
• Lecture 15 Bivariate Longitudinal Data Analysis.
  • Notes 15 (pdf) on bivariate longitudinal data analysis.
  • Examples 15. Two Kenya study analyses (doc).

Labs Labs 2--6 have an assignment to be turned in. Labs are due one week after the day of the lab.

• Lab 1: Introduction to basic data management in SAS for longitudinal data
  • Lab1 (word) Introduction to basic data management in SAS for longitudinal data
  • Text Pain data
  • SAS Pain data
  • SAS Dental Data
• Lab 2: Graphics and Response Profile Models for Longitudinal Data
  • Lab2 using SGplot (word)
  • SAS Lead data
  • SAS Small Mice data
  • Lab2 old using SASgraph(word)
  • Lab2 old SAS graphs using SASgraph
• Lab 3: Pattern covariance models. Mean modeling.
  • Lab3 (word)
  • SAS Lead data
  • SAS Pain data
• Lab 4: Random Effects models. Choosing a mean model.
  • Lab4 (word)
  • Lab4 graphs (word)
  • SAS Dental data
  • SAS CD4 data. See page 228 in ALA book for information on this data set.
  • Lab4 (word) using old sasgraph
  • Lab4 graphs (word) from old sasgraph
• Lab 5: Glimmix: Logistic and Poisson random effects models.
  • Lab 5 (word)
  • Contraception data (text). See ALA book page 415 for information.
  • Epilepsy data (SAS). See ALA book pages 9 and 421 for information
• Lab 6: Genmod for Binary and Count Data
  • Lab 6 sgplot (word)
  • Lab 6 graphs
  • Contraception data (text). See ALA book page 415 for information.
  • Epilepsy data (SAS). See ALA book pages 9 and 421 for information
  • Lab 6 (word) old sas graph
• Lab 7: Hierarchical Models
  • Lab 7 (word)
  • Television School and Family Smoking Prevention and Cessation Project from chapter 22 (text). See ALA 2nd ed book page 639 for additional information.
• Lab 8
  • Lab 8 (word)
  • Lab 8 comments (word)
  • Pain Data (text)