Table A6:
Attrition Corrected Treatment Effects on Factor Scores
| Adjusted Beta | Attrition bounds | ||
|---|---|---|---|
| Unweighted (1) |
IPW (2) |
95% CI (3) |
|
| Maternal Factor Scores | |||
| Mental Health (6m) | 0.205*** (0.052) |
0.205*** (0.052) |
[−0.021 0.394] |
| Mental Health (12m) | 0.170*** (0.054) |
0.169*** (0.054) |
[−0.105 0.385] |
| Mental Health (24m) | −0.002 (0.057) |
0.001 (0.058) |
[−0.294 0.231] |
| Mental Health (36m) | 0.268*** (0.078) |
0.265*** (0.077) |
[−0.159 0.372] |
| Functioning (6m) | 0.182** (0.075) |
0.182** (0.075) |
[−0.052 0.364] |
| Functioning (12m) | 0.195*** (0.069) |
0.196*** (0.069) |
[−0.027 0.406] |
| Functioning (24m) | −0.036 (0.072) |
−0.034 (0.074) |
[−0.341 0.204] |
| Functioning (36m) | 0.287***
(0.081) |
0.280*** (0.079) |
[−0.200 0.349] |
| Child Factor Scores | |||
| Physical Health (6m) | −0.021 (0.079) |
−0.021 (0.079) |
[−0.217 0.204] |
| Physical Health (12m) | 0.019 (0.070) |
0.021 (0.069) |
[−0.187 0.287] |
| Physical Health (24m) | −0.100 (0.083) |
−0.091 (0.083) |
[−0.404 0.171] |
| Physical Health (36m) | −0.166* (0.088) |
−0.160* (0.089) |
[−0.434 0.154] |
| Socioemotional Skills (6m) | 0.187*** (0.056) |
0.187*** (0.056) |
[−0.006 0.375] |
| Socioemotional Skills (12m) | 0.389*** (0.070) |
0.389*** (0.070) |
[0.149 0.601] |
| Socioemotional Skills (24m) | −0.065 (0.067) |
−0.060 (0.071) |
[−0.347 0.162] |
| Socioemotional Skills (36m) | 0.063 (0.075) |
0.073 (0.076) |
[−0.308 0.291] |
| Cognition (12m) | −0.080 (0.083) |
−0.080 (0.084) |
[−0.357 0.227] |
| Cognition (36m) | 0.065 (0.075) |
0.065 (0.073) |
[−0.234 0.364] |
| Investment Factor Scores | |||
| Parental Investment (12m) | 0.075 (0.086) |
0.075 (0.087) |
[−0.204 0.309] |
| Parental Investment (36m) | 0.111 (0.076) |
0.111 (0.078) |
[−0.274 0.321] |
Column 1 reproduces the adjusted treatment effect coefficient on factor scores without attrition correction. Column 2 reports attrition-corrected treatment effect estimates using Inverse Probability of Attrition Weighting (IPAW), where the weights are the inverse of the predicted probability to participate in the respective follow-up based on the baseline covariates. Column 3 shows 95 percent confidence intervals for the treatment effect using attrition bounds that are tightened by child gender and the indicator of whether the baseline PHQ-9 score is above the median (Lee, 2009).
Note:
*
p<0.1;
**
p<0.05;
***
p<0.01