12 Apr module 02: multilevel analysis
- module 02: multilevel analysis
07/09/2020 - 11/09/2020
9:30 am - 11:00 am
Mon 07/09: 9:30 am - 12:45 pm | online
Tue 08/09: 9:30 am - 12:45 pm | online
Wed 09/09: 9:30 am - 12:45 pm | online
Thu 10/09: 9:30 am - 12:45 pm | online
Fri 11/09: 9:30 am - 11:00 am | online
week 1 | register now
Multilevel analysis – also called mixed model analysis – deals with data where not all observations are independent. This includes data generated through longitudinal studies, where the same individual is measured repeatedly over time, grouped data, where individuals within a group are more similar to one another compared to people outside the group, or laboratory tests where the experiments were carried out in separate batches, and where batch effects cannot be excluded. Starting from ANOVA models with random factor levels, the concepts of mixed models are introduced and the basics about inference in random-effects models will be explained. Afterwards, the mixed ANOVA model is extended to general linear mixed models for continuous data. Omitting as much theoretical details as possible, sufficient background will be given such that practicing statisticians can apply mixed models in a variety of contexts, know how to use up-to-date software, and are able to correctly interpret generated outputs. Many applications, taken from various disciplines, will be discussed.
At the end of this module, participants should be able to:
- recognize clustered data
- know what statistical analysis methods should be used to account for the clustered nature
- formulate, fit and interpret basic mixed or multilevel models for the analysis of clustered data, repeated measures or longitudinal data
The module is taught in 9 sessions of 1.5 hours, 4 of which are foreseen as hands-on sessions. You will need a laptop with R (or Rstudio) installed on it.
Participants should have a thorough knowledge of:
- Hypothesis testing
- Linear regression
- Simple linear regression
- Multiple linear regression
- Testing for continuous variables
- Testing for categorical variables
- Confidence intervals
- Analysis of variance
- Basic knowledge of data handling and fitting linear models in R.
If you have no idea what the following commands mean, the course is too advanced for you:
mydata <- read.table (“c:/temp/rawdata.txt”,header=T, dec=”,”)
BACKGROUND READINGSFitzmaurice, G.M., Laird, N.M., and Ware J.H. (2004). Applied Longitudinal Analysis. Hoboken, New Jersey: John Wiley & Sons.
Dr. Ella Roelant obtained a PhD in Mathematics (Statistics) at Ghent University in 2008 on the topic 'Multivariate methods for robust estimation and inference'. Currently she is working as a postdoctoral researcher at StatUa, the core facility of statistical data analysis of Antwerp University and as a statistical consultant at the Antwerp University Hospital. At StatUa her main duty is teaching statistics courses. In the past Ella has taught statistics courses at Ghent University and Newcastle University and worked as a research associate in the Institute of Health and Society at Newcastle University. Her main interest lies in medical statistics.
Prof. Dr. Erik Fransen obtained a PhD in Biochemistry at the University of Antwerp in 1998. After a few years of lab work, he started additional studies in biostatistics, resulting in a Master degree in Statistical Data Analysis at the UGent (2007). Through his research at the Center for Medical Genetics, Erik has experience with a wide range of statistical techniques for data analysis. In addition, he is an experienced teacher of statistics for a non-statistical audience.