Fig. 2

Absolute standardized differences in unweighted and weighted samples [CHE (40% of capacity to pay)]. Results based on separate logistic regressions of insurance status in the follow-on years—2011 and 2013 on the explanatory variables shown in the figure. Balance is assessed by comparing the means of unweighted and weighted treatment and control in the full balanced sample using absolute standardized differences. Balance is said to be achieved if absolute standardized difference is less than the 0.1 threshold (stripped vertical line)