Table 3.
Mechanism check: heterogeneity of treatment effects after 3 and 6 months, by level of payout.
| Wfa (z score) |
Weight (kg) |
Wfa (z score) |
Weight (kg) |
Wfa (z score) |
Weight (kg) |
|||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 3 mo. (1) |
6 mo. (2) |
3 mo. (3) |
6 mo. (4) |
3 mo. (5) |
6 mo. (6) |
3 mo. (7) |
6 mo. (8) |
3 mo. (9) |
6 mo. (10) |
3 mo. (11) |
6 mo. (12) |
|
| Proportional | 0.0703* (0.0416) |
0.0947* (0.0496) |
0.0760 (0.107) |
0.203* (0.121) |
0.0638 (0.0406) |
0.0880* (0.0497) |
0.138 (0.0854) |
0.210** (0.104) |
0.0642 (0.0408) |
0.0836* (0.0481) |
0.138 (0.0858) |
0.202** (0.100) |
| Difference to mean payout of group (×1000) | 0.0111 (0.0129) |
0.0351** (0.0144) |
0.0256 (0.0215) |
0.0403 (0.0280) |
0.0181 (0.0126) |
0.0307** (0.0122) |
0.0342 (0.0273) |
0.0667** (0.0258) |
0.0180 (0.0122) |
0.0278** (0.0123) |
0.0341 (0.0268) |
0.0610** (0.0255) |
| Difference to mean payout*Proportional (×1000) | −0.0101 (0.0146) |
−0.0508*** (0.0170) |
−0.0376 (0.0269) |
−0.0935** (0.0358) |
−0.0158 (0.0140) |
−0.0447*** (0.0139) |
−0.0317 (0.0300) |
−0.0920*** (0.0293) |
−0.0163 (0.0145) |
−0.0377*** (0.0140) |
−0.0316 (0.00310) |
−0.0783*** (0.0289) |
| Child and mother-level controls | X | X | X | X | X | X | X | X | ||||
| Worker Controls | X | X | X | X | ||||||||
| N | 2342 | 2272 | 2348 | 2325 | 1665 | 1934 | 1665 | 1935 | 1665 | 1934 | 1665 | 1935 |
| adj. R-sq | 0.001 | 0.005 | −0.001 | 0.003 | 0.036 | 0.036 | 0.335 | 0.342 | 0.035 | 0.041 | 0.334 | 0.345 |
Notes: Dependent variables are the child's weight-for-age z score (columns 1, 2, 5, 6, 9 and 10), or weight in kilograms (columns 3, 4, 7, 8, 11 and 12). Difference to mean payout refers to each worker's relative position in their treatment group, based on the number of children whose malnutrition status improved at their center relative to other centers in their geographic location and treatment arm as defined by equations (1), (2) in the main text. Columns with odd numbers use outcomes measured at Endline I in April 2015, which is 3 months after the contest started and was the basis for incentives paid in May 2015. Even-numbered columns use outcomes at Endline II in July 2015. Coefficients have been multiplied by 1000 for clarity. All other notes are as for Table 2.