Methods. Randomized control trials were used to conduct evaluations. Multistage randomized sampling methods were used to select 106 intervention schools (41 in three provinces of Afghanistan namely Badakhshan, Ghazni and Takhar and 65 in two provinces of Pakistan namely Punjab and Sindh) compared to 113 control schools (42 in Afghanistan and 71 in Pakistan). Overall, 5603 grade 3 and 5 children (2544 from Pakistan and 2519 from Afghanistan) were sampled and interviewed at baseline in 2018. No significant difference was found at baseline between the children in intervention and control groups in any of the demographic and socioeconomic characteristics, in psychosocial skills or in academic performances. Psychosocial skills were measured using the Multidimensional Scale Life Skills in Late Childhood, the Self-Efficacy Questionnaire and the Child and Youth Resilience Measure. Academic performance was measured using locally defined Mathematics, Languages and General Knowledge tests. We also collected data on stigma and discrimination. a second wave of interviews were conducted in 2020 and a third wave is ongoing. Reliability and validity analysis of all instruments were conducted in both countries at baseline. Analysis of change regression models were used to compare the difference in scales between intervention schools and control schools at baseline and midline.
Results. 1745 of the 2519 children from Afghanistan and 1985 of the 2544 children from Pakistan were still in school in 2020. The rest of children dropped out of school or migrated out of our project areas. Twice more girls than boys dropped out of school between 2018 and 2020. In Pakistan, our interventions significantly improved the life skills of girls by 1.7 point after controlling for age and the districts of the schools (p<0.05). In Afghanistan, our interventions significantly improved self-efficacy by 4.2 point in one rural area (p<0.05). Children in intervention schools were also significantly more aware of stigma and discriminations. Academic performances were not found significantly different between the two groups.
Implications. Our preliminary results indicate that our invention had significantly improved psychosocial skills, especially for girls. Non-significant differences were found in different domains of academic performances probably partially because of the disruption of the Covid-19 pandemic during the long period of lockdown in 2020. To further explore the effects of the intervention on children, a third wave of data will be available current 2022. Multilevel and longitudinal analysis will be used to further identify the mechanism of interventions.