I am working with a mixed linear model where I have several groups, each with a different number of repeated measures. I fit a separate model for each group, but I am facing an issue when it comes to using data from participants with lower counts of repeated measures.
For example, if I fit a model for a group with five repeated measures, I cannot utilize the data from participants who have only three repeated measures. This leads to a loss of valuable information from those participants. Is there a method or approach to handle this issue in mixed linear models, so that I can include all participants and make use of the data with varying numbers of repeated measures across groups?
Another aspect I would like to address is the analysis of errors in the model. I have calculated the absolute error for each observation in the test set. I am wondering if it would be appropriate to use an ANOVA or t-test to check for potential differences in the error for a specific value in a particular feature. Is this a valid approach, or should I consider alternative methods for assessing differences in errors across feature values?
Any guidance or references to best practices in this area would be greatly appreciated.
