Panel A: Institutional Delivery Sample |
---|
Block | Insured | Uninsured | P-Value |
---|
Block 1 | 0.033 | 0.026 | 0.0395 |
Block 2 | 0.066 | 0.062 | 0.1060 |
Block 3 | 0.090 | 0.085 | 0.0305 |
Block 4 | 0.153 | 0.139 | 0.0208 |
Block 5 | 0.280 | 0.269 | 0.4175 |
Block 6 | 0.457 | 0.480 | 0.6198 |
N = 1137 | 89 | 1048 | |
Panel B: Skilled Birth Attendant Sample |
Block | Insured | Uninsured | P-Value |
Block 1 | 0.028 | 0.025 | 0.3297 |
Block 2 | 0.073 | 0.070 | 0.4539 |
Block 3 | 0.135 | 0.138 | 0.5857 |
Block 4 | 0.256 | 0.250 | 0.7342 |
Block 5 | 0.462 | 0.492 | 0.4877 |
N = 1017 | 68 | 949 | |
- Note: The above table reports the test of balance of the mean propensity score between the insured and uninsured by blocks. The results are from implementing “pscore.ado” program in Stata. The p-values are based on a two-sample t-test with equal variance. The pscore command fits a logit (probit is the default) model with a starting specification of linear terms without interactions or higher order terms. If balance in not achieved in a block, the sample in the block is split into equally spaced intervals, with higher order terms and interactions included, and the average propensity score of the treated and controls is re-tested until balance is achieved