2010 Summer Workshop on Longitudinal Data Analysis: June 7-11

This workshop focused on important topics in longitudinal data analysis, with a special emphasis on categorical data. The topics include:

  • Advantages and challenges of longitudinal data analysis
  • Special considerations for analysis such as covariance/correlation structures
  • General approaches to longitudinal analysis including mixed effects models, generalized estimating equations (GEE) models
  • Survival Analysis
  • Growth Curve models
  • Multilevel Growth Trajectories

Computers and software packages: STATA, HLM, and MPlus.

Course Outline

Day 1: General Overview of Longitudinal Analysis (by Rajulton Fernando)

Advantages and challenges of using longitudinal data, data layout and notations, covariance/correlation structures, GEE models for categorical data analysis.

Day 2: Markov Transition Models and Binary Sequence Models (by Rajulton Fernando)

Analysis of turnover tables and binary sequence models. Emphasis is on using models for nominal and ordinal variables, with link functions logit, probit, and complementary log-log.

Day 3: Survival Analysis (by Alain Gagnon)

Discrete-time survival models with various link and risk functions. After introducing basic model building strategies with person-period data, the instructor shows how to account for unobserved heterogeneity and sample selection bias with discrete-time survival data.

Day 4: Growth Curve Models (by Piotr Wilk)

Introduction to structural equation modelling, comparison between hierarchical and structural equation model specification and estimation, unconditional and conditional latent curve models for binary, ordered categorical and count variables.

Day 5: Multilevel Growth Trajectories (by Piotr Wilk)

Introduction to the analysis of longitudinal data using multilevel models for binary and categorical variables. Examples of data preparation are given in STATA and model analysis are demonstrated using HLM.

Instructors

Rajulton Fernando
Associate Professor, Department of Sociology, University of Western Ontario

His research interests are techniques of longitudinal and event history analysis, and modelling demographic phenomena such as fertility, mortality, migration and family life histories. Rajulton and colleagues organized an international workshop on “Longitudinal Research in the Social Sciences: Assessment and Dissemination of Tools for Analysis in the Canadian Context” at Western in 1999, funded by SSHRC’s Research Development Initiative Program. Papers presented in this workshop by international scholars were published as a Special Issue of Canadian Studies in Population in 2001. Rajulton has been giving a course on Longitudinal Data Analysis since 2000, once in every two years.

Alain Gagnon
Assistant Professor, Department of Sociology, with a cross-appointment at the
Department of Epidemiology & Biostatistics
University of Western Ontario

Alain is using historical as well as contemporary data to investigate the demographic and genetic aspects of longevity, focusing on how early life conditions shape health and mortality in later life. He has used survival analyses in much of his recent work on longevity, with a focus on unobserved heterogeneity and sample selection issues. Alain has taught multivariate statistics at both graduate and undergraduate levels since 2003.

Piotr Wilk
Community Health Researcher/Educator, Middlesex-London Health Unit

Piotr research focuses on the health and well being of parents and their children. He is currently conducting research on how the socio-economic conditions in which children are born and grow up affect their health and developmental trajectories. His research also focuses on health of Aboriginal children, examining the role of contextual predictors related to family characteristics and community/neighborhood characteristics. At the Department of Epidemiology, he teaches advanced graduate courses in social statistics and quantitative research methods.

Western provides the best student experience among Canada's leading research-intensive universities.