Examining whether a repeated measure is stable high, or non-stable over time

Dear all,

I have conducted an experiment with a patient versus a control group, in which I asked them to do a reaction time tasks with control words and specific symptom words. The same reaction time task was used on three different days. We want to know whether the patient group shows a bias (i.e. quicker reaction time) to symptom words compared to control words. And we want to know whether this bias is more stable or chronic over time (i.e. less fluctuation of difference in reaction time) for the patient versus control group. I have so far looked at the test-retest reliability for the consecutive days separately for both groups. However, I have also looked at the coefficient of variation per subject, and then calculated the mean difference of this coefficient per group. Finally, I have looked at the stability paths of a SEM cross-lagged panel design. Can someone please explain the exact difference between those three tests (coefficient of variation, test-retest reliability and stability paths in crosslagged panel design). All three tests show very different results, which one of them should I use for my questions. Or does someone have any other suggestions? Thanks for reading!