Missing comparative standard deviations
Posted: Sun Nov 25, 2012 5:57 pm
I am trying to fit longitudinal growth models with measurement occasion at level 1 and individual subject at level 2. The time points in the dataset are unbalanced, with different numbers of measurement occasions and variable time points for each individual included. My problem is that the level 1 variance appears to be a linear or quadratic function of time, but when I fit such models and check the residuals most of the comparative standard deviations at level 1 result in error messages and missing values. However, the other diagnostics for model fit are returned and do not seem to show any major problems. I do not particularly need the comparative SD values at level 1, but I am worried that the errors being returned may indicate something fundamentally wrong with the model being fitted to the data. I have tried looking into this issue, but have not managed to answer the question as to whether these errors can be ignored, or whether they indicate a problem that needs to be addressed.
Any help or guidance on this issue would be very much appreciated!
Many thanks
Oliver
Any help or guidance on this issue would be very much appreciated!
Many thanks
Oliver