Table 4

Treatment Effects Estimated through ANCOVA: Full Sample, and Heterogeneity by Child Gender, Household Poverty, and Geographical Areas

	Girls	Boys	Girls	Boys	Girls	Boys
Panel B: Gender
School feeding	0.242	0.0645	0.205	0.0758	0.273	0.0797
	[0.079 – 0.405]	[–0.116 – 0.245]	[0.042 – 0.368]	[–0.084 – 0.235]	[0.0887 – 0.456]	[–0.107 – 0.267]
Unadjusted p-value	0.004^***	0.478	0.014^**	0.349	0.004^***	0.399
R-W p-value	0.006^***	0.573	0.012^**	0.573
Observations	1,071	1,207	1,067	1,207	1,085	1,229
R-squared	0.089	0.062	0.137	0.135	0.150	0.144
Mean treatment endline	0.158	0.0700	0.120	0.0923	0.237	0.156
Mean control endline	−0.0878	−0.00534	−0.102	−0.0170	−0.0509	0.0440

	Below the Poverty Line	Nonpoor	Below the Poverty Line	Nonpoor	Below the Poverty Line	Nonpoor
Panel C: Household Poverty
School feeding	0.309	0.101	0.233	0.0883	0.328	0.114
	[0.099 – 0.519]	[–0.0642 – 0.267]	[0.037 – 0.428]	[–0.0755 – 0.252]	[0.114 – 0.542]	[–0.072 – 0.299]
Unadjusted p-value	0.004^***	0.227	0.02^**	0.287	0.003^***	0.227
R-W p-value	0.008^***	0.403	0.022^**	0.403
Observations	539	1,739	537	1,737	542	1,772
R-squared	0.090	0.071	0.089	0.151	0.123	0.151
Mean treatment endline	0.136	0.104	0.0704	0.116	0.177	0.199
Mean control endline	−0.186	−5.20e-05	−0.203	−0.0115	−0.177	0.0539

	North	South	North	South	North	South
Panel D: Place of Residence
School feeding	0.253	0.0508	0.243	0.0308	0.297	0.0496
	[–0.005 – 0.511]	[–0.118 – 0.220]	[0.012 – 0.474]	[–0.148 – 0.209]	[0.023 – 0.570]	[–0.146 – 0.245]
Unadjusted p-value	0.054^*	0.549	0.04^**	0.731	0.034^**	0.613
R-W p-value	0.061^*	0.797	0.062^*	0.797
Observations	1,083	1,195	1,087	1,187	1,096	1,218
R-squared	0.043	0.099	0.098	0.163	0.090	0.193
Mean treatment endline	0.179	0.0418	0.169	0.0382	0.270	0.115
Mean control endline	−0.0639	−0.0273	−0.0629	−0.0509	−0.0160	0.0127

Notes:
↵* p < 0.1,
↵** p < 0.05,
↵*** p < 0.01.
Confidence intervals clustered at community level in squared brackets. R-W p-values were adjusted for multiple testing using the Romano–Wolf (2005, 2016) step-down method with 2,000 iterations and standard errors clustered at community level. The table above presents intent-to-treat effects on each outcome estimated through ANCOVA for the full sample and stratified by child gender, household poverty, and place of residence. Models were estimated through OLS. For each outcome, the model controls for its baseline value, a dichotomous variable related to the randomized assignment to school feeding, and region dummies. Math and literacy scores are age-standardized. The composite index of learning was computed as the average of the math and literacy scores, and then they were standardized to the control group within each round. Household poverty is a dichotomous indicator having the value of one if the household had baseline per capita consumption levels falling below the national consumption poverty line in 2013. Northern regions include Upper West, Upper East, and Northern region. Southern regions include Western, Central, Greater Accra, Volta, Eastern, Asanti, and Brong Ahafo.