Cost-effectiveness of two parenting interventions in Côte d’Ivoire

This project is built around 2 clustered Randomized Control Trials (cRCTs).

First RCT
The intervention of PNN (Programme National de Nutrition), with technical support of HKI (Helen Keller International), combines in-group training sessions for children’s caregivers, home visits and community sensibilization, targeting all households with children 0-5 years old in each treated locality. Trainings are organized for groups of 15 caregivers, over an initial period of 10 months (one one-hour session per month), integrating topics on early childhood learning and stimulation, positive discipline and socio-emotional support with nutrition and hygiene messages, adapted from the “Care for Child Development” program. Refresher trainings follow for another 10 months. This is complemented with home visits (1-1.5 hour, once a month for each household), conducted by the same community agent who leads the in-group training, and locality-level sensibilization activities.

Second RCT
The Ministry of Woman, Family, and Child, with support of IRC (International Rescue Committee), provides formal in-group training sessions around the FMD (“Families Make a Difference”) curriculum, and encourages the organization of Communities of Practice (CoP). Formal trainings are organized for 25 caregivers of children 0-5 years old per village, once a week, for 11 consecutive weeks. In the fourth week, 5 beneficiary parents that are particularly active, interested and engaged will be selected to become voluntary parent leaders of a Community of Practice (CoP). Each will identify 3 additional parents with children 0-5 years old from the community (not already participating in the formal trainings). These 5 parent leaders + 15 parents identified by them will form a CoP, meeting together so leaders can share lessons learned during the formal training sessions.

2021-08-01

2023-03-31

Cognitive and socio-emotional development

We will use internationally standardized tests to measure different aspects of cognitive and socio-emotional development, and aggregate them into 2 overall indices.

We will measure impacts on outcomes along the causal chain, including parental knowledge, parental beliefs about the returns to early parental investment, targeted parenting behaviors, practices, and household allocation and investment related to ECD, as well as the primary caregiver’s mental health. We will also measure social norms, second-order beliefs, and social interactions.

First RCT:
The project will take place in all 171 localities in 33 sanitary areas of Lakota district. The sanitary areas are gradually phased in, rolling out activities in blocks of about 15-25 localities (organized by sanitary area) at a time, with a new block starting activities every month for a period of 10 months.
1- For the first 3 blocks of localities, half of the localities are randomly selected as (temporary) controls. This results in 35 control localities, where the start of the intervention will be postponed until all the other sanitary areas have been phased in allowing to estimate the program impacts of parental exposure to in-group training and home visits on behavioral and ECD outcomes after a period of 8-10 months
2- In half of the treatment localities, randomly selected, project staff will identify local (opinion) leaders and ask them to become ECD community champions. These influencers will receive an orientation about the content of the trainings, asked to share concerns and opinions, and asked for advise on how to change ECD practices in the locality, and actively encouraged to engage with and participate in the program.
3- In half of the treatment localities, randomly selected fathers will be shown a set of short videos, specifically created to sensitize fathers about their participation in early childhood learning. The local languages videos (created by DMI) are shown to the fathers of the sampled households by baseline enumerators, with a script encouraging them to participate in the upcoming trainings. Randomization of the father’s exposure to videos will be orthogonal to the selection for the community champions.

All randomizations will be stratified by sanitary area.

2nd RCT
1. FMD trainings will take place in 158 localities, with 103 randomly-selected localities to be reached in year 1, and the remaining 55 in year 2 (serving as the one-year experimental control group). The 158 localities are spread across 5 regions, 9 department and 20 sous-prefectures. The selection of the 25 beneficiary parents to be invited to the training will occur in the 158 localities before the start of trainings in year 1.
2. Grandmothers or other elderly female family members will be invited to separate formal in-group training sessions in 34 randomly selected treatment localities.
3. In 34 randomly selected treatment localities the CoP will not be promoted.

All randomization will be stratified by sous-prefecture (geographical units)

Just after the start of the intervention in year 1, one of the (private sector) partners of the project wanted to roll-out a preschool program (involving preschool construction, curriculum development, teacher training and assignment and parental awareness) in a small subset of the year 1 and year 2 villages. A total of 36 candidate villages for those preschools were identified among the 158 initially targeted schools. After they were identified 15 of the 36 villages were randomly selected to receive the preschool intervention, with 10 randomly selected to receive this intervention starting in 2021, and another 5 randomly selected to receive the intervention starting in 2022. This randomization was stratified by region and by the year 1 versus year 2 parental training assignment.

Not available

randomization done in office by a computer

Randomization at locality level

Yes

RCT1: 171 localities
RCT2: 158 localities

RCT1: 2052 children RCT2: 3634 children (including 948 children to test spillovers to households not receiving main training)

RCT 1: 35 late treatment localities; 35 early treatment localities; 68 localities with community champions; 68 localities without community champions; 68 localities with videos to fathers; 68 localities without videos to fathers.

RCT2: 55 control localities; 103 treatment localities; 69 localities with extra grandmother sessions; 34 localities without grandmother sessions; 34 localities without CoP; 69 localities with CoP;

For design of the first RCT allows to detect a MDE of .25 standard deviations (sd) for the overall treatment effect, and .18 sd for the comparing treatment arms (community champions, and videos-to-fathers). The design of the second RCT has a MDE of 0.19 sd for overall impacts, and 0.23 sd for differences of treatment variations (CoP, grandmother training), and 0.3 sd for effects on ECD beliefs and behavior households possibly indirectly exposed to training through the CoP.