The dataset (sample, panel) changes month to month and week to week. How do I account for this in my longitudinal analysis?