I\'m creating a Bayesian multilevel model in rstanarm to predict satisfaction by party identity in each year (using year as the second level), with demographic controls. (I\