Table 4. Average treatment effect on the treated of insurance on the utilization of facility-based delivery using propensity score weighting or matching, The Philippines.
| Alternative Matching Methods | Propensity Score Weighted | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Propensity Score | Nearest Neighbour | Kernel-PS matching | Ordered Probit, unadjusted | Ordered Probit, adjusted | ||||||
| ATT | SE | ATT | SE | ATT | SE | ATT | SE | ATT | SE | |
| All | 0.0973*** | (0.035) | 0.0547* | (0.028) | 0.0786** | (0.032) | 0.07** | (0.031) | 0.0802*** | (0.026) |
| Locality | ||||||||||
| Rural (N = 816) | 0.1127** | (0.051) | 0.0873** | (0.037) | 0.0964** | (0.039) | 0.0905** | (0.038) | 0.091*** | (0.032) |
| Urban (N = 560) | 0.0486** | (0.025) | 0.0764 | (0.048) | 0.0648 | (0.054) | 0.0622 | (0.051) | 0.0473 | (0.036) |
| Wealth | ||||||||||
| Poor (N = 623) | 0.1102** | (0.046) | 0.0868* | (0.046) | 0.087 | (0.053) | 0.0936** | (0.042) | 0.1011** | (0.039) |
| Non-poor (N = 729) | 0.053 | (0.035) | 0.0576 | (0.037) | 0.0683* | (0.04) | 0.064* | (0.034) | 0.0614** | (0.029) |
Notes: Dependent variable is facility-based delivery. Standard errors are in parentheses. For parametric specifications, they are clustered at the primary sampling unit. For PS and NN matching, Abadie and Imbens (2006, 2011, 2012) derived standard errors are used. For Kernel-PS matching, bootstrapped standard errors using 1,000 replications of the sample are used.
(*), (**), and (***) denote significance at 10%, 5%, and 1% levels, respectively. ATT, average treatment effect on the treated; S.E., standard error; PS, propensity score; NN, nearest neighbour.