From: Comparison of methods for handling covariate missingness in propensity score estimation with a binary exposure
Group
x6
I{Y > mean(Y)}
% of observations
% missing on x2
1
0
30
25
50
2
40
60
3
20
4
10
15
Total