Course: Applied Longitudinal Data Analysis: modelling change over time

Frequency:
Once per year

Content and aim
This course focuses on the analysis of longitudinal data with continuous outcome variables. By extending the well-known multiple linear regression model step by step, we will be developing longitudinal data analysis techniques for studies with repeated observations on the same respondents. This will result in the introduction of the linear mixed effects model (also known as multilevel model, random-effects model, hierarchical linear model, ...), allowing the analysis of change over time (such as change of test-scores over time, growth of any kind).

Throughout the course lectures, the emphasis will be on understanding the why and how of these models by explaining the underlying theory of these multilevel analyses using lots of practical examples. The application of these techniques will be demonstrated in both SPSS and R. The lectures will follow topics and theory along the lines of the first half of the book Applied Longitudinal Data Analysis by J.D. Singer and J.B. Willet (2003, ISBN-139780195152968).

Learning outcomes

• Explore (graphically and numerically) longitudinal data with continuous outcome and recognize the need for mixed effects models (multilevel models)
• Understand the theory behind the multilevel model for change
• Build, examine, interpret, expand and compare linear mixed effects models
• Perform all described techniques using either SPSS or R 

Organization and materials used
Within two weeks, five lectures of 2- 2.5 hours each will be given. Each lecture is followed by  a workshop, in which practical exercises will be provided to be performed using either SPSS or R. Lecture slides, exercises and worked-out answers for both SPSS and R will be provided online. Additional reading material is the book Applied Longitudinal Data Analysis by J.D. Singer and J.B. Willet (2003, ISBN-139780195152968). In the book and online material which accompany this course, scripts and data for the examples can also be found for self-study using other software packages such as SAS and Stata (and in lesser extent: HLM, MLwiN and Mplus).

Intended for
PhD students and master students. Students entering this course should have firm knowledge of and experience in applying basic statistical concepts and theory, including multiple linear regression analysis. This basic knowledge is provided by the Basic Medical Statistics course.

Note: for more in-depth model building, techniques for non-continuous outcome variables, the use of Generalized Linear Mixed models and GEE, we recommend the course Generalized Linear and Mixed Effects Models.

Requirements     
Basic Medical Statistics (or an equivalent course, see above)

EC (without exam)

1.5

EC (with exam)

2

Course coordinator

• Sacha la Bastide, PhD
  

Language

English

Contact person

Back to listing