Evaluating Parenting Home Visiting Program and Unconditional Cash Transfer in Thailand

The main intervention in this project is the Jamaican Home Visiting program, so-called Reach Up (Grantham-McGregor et al., 1991). This program aims to facilitate developmentally appropriate learning activities between mothers (primary caregivers) and
their children by demonstrating play activities centered around daily routines. Most of picture books, naming plates, conversation scenes, puzzles, lotteries, and blocks were prepared by the project, which is different from the original Reach Up program, where most of them were drawn on resources in the home. Materials for the stimulation component were adapted from the Jamaican intervention
to fit with the local cultural context and to the average educational level of home visitors and
program beneficiaries.

Play activities followed steps aimed towards children’s gradual mastery
of a learning objective: (1) child observes the play activity (modeling), (2) mother and child
do the activity together, (3) child attempts the play activity on his/her own, (4) mother
prompts naming and verbalization of objects and actions linked to the play activity, (5) the
developmental level of the play activity is adjusted by increasing or decreasing the challenge
based on the child’s performance (scaffolding). Throughout the play activities, mothers are
encouraged to provide children with contingent positive reinforcement for progress toward
the learning goals (praise) and to follow the child’s interest.

Similar to the Colombian project (Attanasio et al., 2020), this project recruited local volunteers as home visitors. Home visits were on a weekly basis and lasted approximately one hour. About 40% of children had a primary caregiver who was not a biological parent. This is similar to the situation in the China Reach project (Heckman et al., 2020). This is a unique challenge for East Asian countries. So, home visits took place with the child and his/her primary caregiver, who may not be a biological parent.

2022-12-01

2023-10-31

Child development measuring by DENVER II and Picture Memory test: this part is also a direct assessment. It is similar to digit span memory and is taken from the Wechsler Preschool & Primary Scale of Intelligence (WPPSI), both of which will be assessed at both baseline and endline surveys.

1. Socio-emotional skills based on Strengths and Difficulty Questionnaire (SDQ).
2. Parental investment including activity time and learning materials.

We employ a clustered randomized controlled trial technique to measure the impact of the intervention. We will randomly select subdistricts in each area into a control group and two treatment groups:
1. Control (C): receive no intervention
2. Treatment 1 (T1): receive the Jamaican Home Visiting program (Reach Up).
3. Treatment 2 (T2): receive a monetary transfer of 4,000 Baht (about 111 USD)

This project will be implemented in 6 areas across the countries, including Chiang Mai in the north (purple), Khon Kean in the northeast (orange), Mahasarakham and Kalasin in the northeast (green), Lop Buri in the central (red), Nakhon Nayok in the central (yellow), and Pattaluk and Songkla in the south (blue). See the figure below. Note that the northeast is the poorest region while the south is where a significant fraction of the population is Muslim.

In each area, we will choose 12-15 subdistricts, each of which consists of several villages. Each subdistrict has at least one childcare center, which usually consists of two levels of classes: (1) class of younger children (2-3 years of age at the beginning of the academic year, 16th of May) and (2) class of older children (4-3 years of age).

The randomization unit is a subdistrict. We will first rank subdistricts by the number of potential samples in the subdistrict and then form groups of three based on the ranking. We randomly assigned each group to the control and two treatment groups.

Randomization was done in the office using MS Excel.

Subdistricts

Yes

81 subdistricts

About 1,400 children

There are 27 subdistricts for each arm

This calculation has been completed before the baseline survey. We therefore assume that each subdistrict should have roughly 20 children. We also assume that the intra-correlation coefficient is 0.05 and control variables would lead to an R-square of 0.15. The targeted significance level is 0.05. We expect an effect size of about 0.25 SD. The power calculation (using the OD program with the targeted power of 0.80) suggested that each arm should have at least 24 subdistricts.