Table A2.
Attrition analysis: Prediction of panel inclusion using treatment status (Panel A), background characteristics, and interactions between background characteristics and treatment (Panel B)
| Malawi [Social Cash Transfer Program] | Zambia [Multiple Category Targeted Grant] | |||||
|---|---|---|---|---|---|---|
| Male | Female | Male | Female | |||
| Panel inclusion | Panel inclusion | Panel inclusion | Panel inclusion | |||
| Panel A | Sample: Ever married or cohabited |
Sample: Ever married or cohabited |
Sample: Ever pregnant |
Sample: Ever married or cohabited |
Sample: Ever married or cohabited |
Sample: Ever pregnant |
| Treatment status | −0.075 | −0.046 | −0.041 | −0.065 | −0.057 | −0.029 |
| (1.62) | (1.26) | (1.44) | (2.56)** | (2.08)** | (1.13) | |
| R-squared | 0.01 | 0.00 | 0.00 | 0.01 | 0.00 | 0.00 |
| Observations | 1,399 | 1,204 | 1,204 | 1,725 | 1,503 | 1,503 |
| Attrition rate | 0.288 | 0.283 | 0.248 | 0.249 | 0.288 | 0.195 |
| Panel B | ||||||
| Treatment status | 0.031 | −0.165 | 0.278 | −0.184 | −0.125 | −0.448 |
| (0.10) | (0.59) | (1.01) | (0.74) | (0.49) | (1.67)* | |
| Baseline characteristics and safe transition indicator | Yes | Yes | Yes | Yes | Yes | Yes |
| Interactions of treatment and baseline characteristics/safe transition indicator | Yes | Yes | Yes | Yes | Yes | Yes |
| R-squared | 0.09 | 0.05 | 0.09 | 0.07 | 0.09 | 0.05 |
| Observations | 1,399 | 1,204 | 1,202 | 1,725 | 1,503 | 1,503 |
| p-value from F-test that Treatment and interactions with Treatment are jointly 0 | 0.008 | 0.044 | 0.008 | 0.005 | 0.212 | 0.476 |
NOTES: Estimations use Linear Probability Modeling. Robust standard errors in parentheses corrected for clustering.
p<0.01,
p<0.05,
p<0.1.
The analysis utilizes baseline data, thus all controls (and interactions) are measured at baseline. Sampling weights have been Malawi. The baseline characteristics included in Panel B are the same as the controls used in our main adjusted specifications, nan education (whether or not currently attending school) of the youth, characteristics of the main survey respondent or transfer recipie whether they have ever attended school and marital status indicators), household size, a dummy capturing whether per capita consi below the mean and an indicator that captures whether the youth lives in a gender-progressive (top half of score) community as wel indicators used for the randomization (traditional authorities dummies in Malawi and districts in Zambia).