The statistical
significance of these adjusted
differences as shown in Figures 2 and 3 are estimated by models that take into account the entire distribution
of responses (
e.g., very satisfied, somewhat satisfied, somewhat dissatisfied, or very dissatisfied).