Whose Variance is it Anyway? Interpreting Empirical Models with State-Level Data