Greetings
I have a measurements on grouped individuals. There are approximately 100 groups, and approximately 500 individuals per group, with measurements taken over 5 years. There are no repeated measurements. The data are observational. The outcome is binary and there are 5 candidate covariates.
Interest is centered on the change in variation of the group level effects. In particular there is a hypothesis that there has been a meaningful decreasing trend in variation of the group effects over the 5 year period.
To investigate this, I'm considering these approaches
1. Slice up the data into years, and run a 2-level linear mixed effects model on each slice;
2. Add a covariate for year (eg 1=2000, 2=2001, 3=2002 etc) and run the model on the whole dataset;
3. Use a 3-level linear mixed effects model: individuals within years within groups.
I would be grateful for any comments on these and/or any alternative approaches.
Thanks
LR
I have a measurements on grouped individuals. There are approximately 100 groups, and approximately 500 individuals per group, with measurements taken over 5 years. There are no repeated measurements. The data are observational. The outcome is binary and there are 5 candidate covariates.
Interest is centered on the change in variation of the group level effects. In particular there is a hypothesis that there has been a meaningful decreasing trend in variation of the group effects over the 5 year period.
To investigate this, I'm considering these approaches
1. Slice up the data into years, and run a 2-level linear mixed effects model on each slice;
2. Add a covariate for year (eg 1=2000, 2=2001, 3=2002 etc) and run the model on the whole dataset;
3. Use a 3-level linear mixed effects model: individuals within years within groups.
I would be grateful for any comments on these and/or any alternative approaches.
Thanks
LR
Last edited: