Mixed Model and Multiple Comparisons

IndependentHare

My study looked at whether begging calls differed between 5 bird species. After recording each individual, I chose 5-10 begging calls to measure from the individual. Thus, I am using a Mixed Model with ID as my random effect and Species as my fixed effect. I need to run multiple comparisons to determine which pairs of species' calls differed. The JMP manual for running multiple comparisons in a Mixed Model states that the Least Squares Means plot only includes the fixed effect portion of the model. Does this mean the Least Squares Means plot I am seeing does not incorporate my random effect "ID" into the plot?

MRB3855

Hi @IndependentHare : I don't know exactly what your model is (beyond the general "mixed" descriptor). But. generally speaking, the answer to your question "Does this mean the Least Squares Means plot I am seeing does not incorporate my random effect "ID" into the plot?" is "no". The random effects are incorporated via incorporatiing/accounting for/etc. varibility between IDs. Again, it will depend on your model, but generally speaking the standard errors (and therefore confidence intervals) for your treatment comparisons account for variability between IDs. You can see this via making your ID factor fixed (or ignore ID completely) and comparing the results with your mixed model results.

 

And welcome to the community!