Book contents
- Applied Longitudinal Data Analysis for Medical Science
- Applied Longitudinal Data Analysis for Medical Science
- Copyright page
- Dedication
- Content
- Preface
- Acknowledgements
- Chapter 1 Introduction
- Chapter 2 Continuous Outcome Variables
- Chapter 3 Continuous Outcome Variables: Regression-based Methods
- Chapter 4 The Modelling of Time
- Chapter 5 Models to Disentangle the Between- and Within-subjects Relationship
- Chapter 6 Causality in Observational Longitudinal Studies
- Chapter 7 Dichotomous Outcome Variables
- Chapter 8 Categorical and Count Outcome Variables
- Chapter 9 Outcome Variables with Floor or Ceiling Effects
- Chapter 10 Analysis of Longitudinal Intervention Studies
- Chapter 11 Missing Data in Longitudinal Studies
- Chapter 12 Sample Size Calculations
- Chapter 13 Software for Longitudinal Data Analysis
- References
- Index
Chapter 13 - Software for Longitudinal Data Analysis
Published online by Cambridge University Press: 20 April 2023
- Applied Longitudinal Data Analysis for Medical Science
- Applied Longitudinal Data Analysis for Medical Science
- Copyright page
- Dedication
- Content
- Preface
- Acknowledgements
- Chapter 1 Introduction
- Chapter 2 Continuous Outcome Variables
- Chapter 3 Continuous Outcome Variables: Regression-based Methods
- Chapter 4 The Modelling of Time
- Chapter 5 Models to Disentangle the Between- and Within-subjects Relationship
- Chapter 6 Causality in Observational Longitudinal Studies
- Chapter 7 Dichotomous Outcome Variables
- Chapter 8 Categorical and Count Outcome Variables
- Chapter 9 Outcome Variables with Floor or Ceiling Effects
- Chapter 10 Analysis of Longitudinal Intervention Studies
- Chapter 11 Missing Data in Longitudinal Studies
- Chapter 12 Sample Size Calculations
- Chapter 13 Software for Longitudinal Data Analysis
- References
- Index
Summary
In Chapter 13, a few examples which were evaluated in the preceding chapters of this book are reanalysed with different software programmes such as SAS, R and SPSS. For all examples, both output and syntax is provided. The most important conclusion of all the analyses is that mixed model analysis and GEE analysis with a continuous outcome variable are very stable and lead to the same results, independent of the software programme used. This also holds for logistic GEE analysis, but is totally different for logistic mixed model analysis. The results of the logistic mixed model analysis are highly dependent on the estimation procedure used. The most important estimation procedures used are the Gauss-Hermite quadrature method and the residual pseudo-likelihood method. It is argued that the Gauss-Hermite estimation procedure provides the most valid results of a logistic mixed model analysis.
Keywords
- Type
- Chapter
- Information
- Applied Longitudinal Data Analysis for Medical ScienceA Practical Guide, pp. 220 - 242Publisher: Cambridge University PressPrint publication year: 2023