Longitudinal Data Analysis with Discrete and Continuous Responses

This course is for scientists and analysts who want to analyze observational data collected over time. It is not for SAS users who have collected data in a complicated experimental design; they should take the Modèles Mixtes : Théorie et applications aux données de santé course instead.

Apprendre à

• create individual and group profile plots and sample variograms
• use PROC MIXED to fit a general linear mixed model and a random coefficient model
• plot information criteria for models with selected covariance structures
• generate diagnostic plots in PROC MIXED
• fit a binary or ordinal GEE model in PROC GENMOD
• compute GEE deletion diagnostic statistics in PROC GENMOD
• fit a generalized linear mixed model in PROC GLIMMIX.

A qui s’adresse cette formation ?

Epidemiologists, social scientists, physical scientists, and business analysts

Formats de formation disponibles	Durée standard (peut varier selon les sessions. Se référer au calendrier)
Formations présentielles:	3 jours
Formations à distance:	6 sessions d'une demi-journée chacune
e-Learning:	21 heures/180 jour licence

Before attending this course, you should be able to

• execute SAS programs and create SAS data sets. You can gain this experience by completing the Programmation SAS I : les fondamentaux course.
• fit models using the GLM and REG procedures in SAS/STAT software. You can gain this experience by completing the Analyse statistique - Niveau 2 course.

Cette formation concerne SAS/STAT logiciel

Longitudinal Data Analysis Concepts

• understanding the merits and analytical problems associated with longitudinal data analysis
• examining the differences between PROC GLM and PROC MIXED

Exploratory Data Analysis

• graphing individual and group profiles
• identifying cross-sectional and longitudinal patterns

General Linear Mixed Model

• understanding the concepts behind the linear mixed model
• examining the different covariance structures available in PROC MIXED
• fitting a general linear mixed model in PROC MIXED

Evaluating Covariance Structures

• creating a sample variogram that illustrates the error components in your model
• plotting information criteria for models with selected covariance structures

Model Development, Interpretation, and Assessment

• learning the model building strategies in PROC MIXED
• creating interaction plots
• specifying heterogeneity in the covariance structure
• computing predictions using EBLUPs
• fitting a random coefficient model in PROC MIXED
• generating diagnostic plots in PROC MIXED using ODS Graphics

Longitudinal Models with Binary Responses

• examining the correlation structures available in PROC GENMOD
• fitting a binary GEE model in PROC GENMOD

Longitudinal Models with Ordinal Responses

• examining cumulative logit plots
• fitting an ordinal GEE model in PROC GENMOD

GEE Model Diagnostics

• computing GEE deletion diagnostic statistics in PROC GENMOD

Generalized Linear Mixed Models

• fitting a longitudinal model with a discrete response using the GLIMMIX procedure