Good morning! I am hoping someone will be able to answer these questions.
1) I am trying to do mixed effects modelling and have recently learned that when I use the 'mixed' command and then try to get predicted means (and plots of those means) using the margins command, the predicted scores only incorporate data from the fixed (but not the random) part of my model. I am able to re-run the analysis using the 'meglm' command instead of 'mixed' and get the same model, but can no longer predict mean scores/margins or plot the data to see how it looks. Can someone please help with this?
2) A second question I have is about comparing two models to see which is a better fit.
I am trying to develop growth curves, and so my first model has a quadratic term in the fixed, but not the random part of the model - megml score age age2 || id: age
The second model has a quadratic term in both the fixed and random parts of the model - megml score age age2 || id: age age2
Can I compare the fit for these two models using AIC/BIC?
3) Is the method for looking at level 1 and level 2 residuals the same if I use megml instead of mixed?
i.e., for Level 1 residuals - predict rlevel1, residual
For Level 2 residuals - predict uslope, uintercept, reffects
I would very much appreciate any advice and thank you greatly in advance for your time!
BJ Cunningham
McMaster University
Hamilton, ON Canada
1) I am trying to do mixed effects modelling and have recently learned that when I use the 'mixed' command and then try to get predicted means (and plots of those means) using the margins command, the predicted scores only incorporate data from the fixed (but not the random) part of my model. I am able to re-run the analysis using the 'meglm' command instead of 'mixed' and get the same model, but can no longer predict mean scores/margins or plot the data to see how it looks. Can someone please help with this?
2) A second question I have is about comparing two models to see which is a better fit.
I am trying to develop growth curves, and so my first model has a quadratic term in the fixed, but not the random part of the model - megml score age age2 || id: age
The second model has a quadratic term in both the fixed and random parts of the model - megml score age age2 || id: age age2
Can I compare the fit for these two models using AIC/BIC?
3) Is the method for looking at level 1 and level 2 residuals the same if I use megml instead of mixed?
i.e., for Level 1 residuals - predict rlevel1, residual
For Level 2 residuals - predict uslope, uintercept, reffects
I would very much appreciate any advice and thank you greatly in advance for your time!
BJ Cunningham
McMaster University
Hamilton, ON Canada
