A Randomized Evaluation of the Effects of a Preschool Intervention on Child Learning and Women's Economic Empowerment in Tharaka Nithi County, Kenya

We refer to the intervention described here as the Tharaka Nithi County Preschool Program.

The trial aims to investigate how a well-designed preschool program can improve child development and school readiness, while at the same time improving women's empowerment through improved employment and earnings and better recognition of their roles as mothers.
Currently preschools in Tharaka Nithi County, Kenya only admit children from 4 years old. And while there is a preschool curriculum for their education, schools tend to be under-equipped with learning materials and the quality of instruction is low.

The intervention consists of six main components: A. Expanding admission to 3-year-olds in the current preschools (they currently start at 4), B. Adapting the government curriculum to accommodate the 3-year-old children (as the curriculum currently starts at 4 years) C. Developing teacher activity guides to support curriculum implementation suitable for both 3- and 4-year-olds, D. Providing developmentally appropriate teaching and learning materials (e.g. picture books, puzzles, games), E. Adding an extra teacher in the preschool class (PP1), and F. Training the teachers how to use the activity guides and learning materials how to use evidence-based and developmentally-appropriate instructional strategies through teacher-training workshops and in-class coaching.

A group of specially trained mentors will train the teachers and will oversee the implementation of the intervention, offering support.

The intervention will last for two successive school years.
Intervention Start Date
2024-01-15
Intervention End Date
2025-10-24

2024-01-15

2025-11-30

Primary Outcomes (end points)
Child Development Outcomes:
• Measurement: Cognitive, linguistic Cognition, Language, School Readiness skills and Behavior and socio-emotional development assessed through appropriate instruments (e.g., IDELA, WPPSI). the Weschler Preschool and Primary Scale of Intelligence (WPPSI), the International Development and Early Learning Assessment (IDELA) and the Preschool Self-Regulation Assessment (PSRA) rating scale.
• Focus: Impact of the innovative preschool intervention on child development outcomes IQ and school readiness in 3-year-olds after one and two years of the intervention, as well as on 4-year-olds after one year compared to control schools.

Women's Labor Supply, Well-being, and Empowerment:

• Measurement: Mother and father surveys capturing labor supply, earnings, total household earnings, entrepreneurial activities, mental health indicators, measures of empowerment, control of resources, and decision-making.
• Focus: Impact of the preschool intervention on women's labor supply, time-use, wellbeing, and empowerment.

1. Quality of the Classroom Environment:

Measurement: Includes three subscales: implementation of learning activities, classroom interactions and use of learning materials.
Focus: Understanding the effectiveness of the intervention in promoting a high quality teaching and learning environment.


2. Cost-Benefit Analysis:

Measurement: Conducting a cost-benefit analysis of the program.
Focus: Assessing the economic viability of the intervention and its potential for scalability.

3. Fidelity of Intervention:

Measurement: Monitoring the fidelity of the intervention, attendance rates of children, and parent knowledge on child development.
Focus: Understanding the implementation process and identifying mechanisms and obstacles towards improved women's empowerment and child development.

4. Teacher Satisfaction
Measurement: Teachers’ satisfaction with their workload, work environment, appreciation from parents and social status in the community.
Focus: Understanding how the intervention affected teachers’ satisfaction with their role.

Experimental Design
The experimental design involves a Randomized Controlled Trial (RCT) to assess the impact of the Tharaka Nithi County Preschool Program on child development and women's labor supply. The intervention targets 3–4-year-olds, utilizing the existing infrastructure of PP1 (pre-primary 1) in selected public schools. The study duration is planned for two years, starting in the school calendar year beginning January 2023.

There will be two experimental cohorts. Cohort 1 recruited in January 2024, following a census on October-December 2023 and Cohort 2 recruited in September - November 2024. The treatment group of Cohort 1 will be subject to the intervention from January 2024 to October 2024, while the treatment group cohort 2 will be subject to intervention from January 2025 to October 2025. data collection will follow the end of the intervention respectively for both cohorts.


There are 149 potentially eligible schools, based on capacity to receive new children and on the fact that they do not include playgroups or regular attendance of three year olds. The census determined that 118 include enough eligible families within their catchment areas. Of these 60 were allocated to treatment and 58 to control.

Cohort 1: Randomization will take place by computer following a census of eligible households (those with children 2y8mo-3y6mo at baseline who have expressed willingness to participate) within a radius of 3km from the 118 schools. Within the 3km catchment area of the schools 15 children will be selected and their families consented for the intervention and data collection. Following this, within the catchment group of the 60 treatment schools 5 children are randomly admitted to the treatment schools and another 5 are randomly selected as controls within the same communities. Within the catchment area of the control schools 5 children and their families are randomly selected for observation. Data is collected from the families (mother and father - when present), and the children (assessments) at baseline, midline, and at the end of the intervention. In addition, within each school (treatment and control) 5 4-year olds are recruited. They are also subject to the intervention, since they are in the same classrooms as the 3-year-olds. However, the non-treated 4-year-olds are still attending PP1 class but in a non-treated school

Cohort 2: The same treatment and control schools are involved. Only 3 year olds are recruited (2y8mo-3y6mo at baseline). We identify the 5 closest consenting families with 3-year-olds to each of the treatment schools. If it is a treatment school, the 5 children are admitted to the PP1 in that school. The five children and their families in the radius of the control school are kept for observation. Consenting takes place before treatment status is revealed. Data is collected from the families (mother and father - when present), and the children (assessments) at baseline and at the end of the intervention.


Preschool teacher interviews and classroom observations will be conducted after 1 year of intervention and again at the end of the second year of intervention in all study schools.


We will select 118 schools in Tharaka Nithi county in Kenya based on: a) their ability to accommodate 5 extra children in their preschool class (PP1 - currently for four year-olds only); b) that do not have a playgroup for children younger than 4; c) do not admit regularly 3-year-olds.

Determining the 3-year olds that will participate in the study: Following a census we consider all 3-year-olds defined as children who will be between 2y8mo-3y6mo in January 2024. We then allocate 10 children and 5 reserve children to the closest possible school of the 118 study schools such that each school is associated with 10 children and five reserve children to fill in the place of those who turn down participation in the study.

We then randomly select 60 schools into treatment. For each treatment school we randomly select 5 children from the group associated with the school (as above) to attend to attend the school. The other five will be controls. Children associated with control schools are also controls. Once parents have consented, these children are assessed, and household data is collected.


4-year-olds attend preschool (PP1) anyway: This is the current age of first attendance. Thus, once school has started in January 2024 we obtain the register of all 4-year-old children in all 118 study schools. We then randomly pick 5 4-year-olds from the list to be the children included in the study. Once parents have consented, these children are assessed, and household data is collected.

Cohort 1: Randomization will take place by computer following a census of eligible households (those with children 2y8mo-4y5Mo 3y6mo at baseline who have expressed willingness to participate) within a radius of 3km from 149 pre-selected schools. The census will be used to determine the final 118 schools to participate in the study and their respective catchment areas (see above).

Cohort 2: The treatment and control schools remain unchanged. Only three-year-olds are recruited. the five closest families with a 3-year-old who consent to the intervention and the data collection are recruited before randomization is revealed to them. Children in the catchment area of a treatment school are then offered admission.

Cohort 1: School and individual.

Cohort 2: School

Yes

118 schools arrived at as follows: 149 eligible schools. Eligibility is based on capacity to admit extra pupils, and not including a playgroup or otherwise admitting 2-year-olds. Further selection based on availability of three-year-olds in catchment area leads to an experimental population of 118 schools.

Cohort 1 recruited in January 2024: 600 4 year-olds and 900 3-year -olds Cohort 2 Recruited in November 2024: 590 3-year-olds

Sample size: planned number of observations
Cohort 1 recruited in January 2024: 600 4 year-olds and 900 3-year-olds Cohort 2 Recruited in November 2024: 590 3-year-olds
Sample size (or number of clusters) by treatment arms
Cohort 1:

60 treatment schools each with: five randomly selected 4-year-olds (treated), 5 three-year-olds randomly selected to attend (treated) and 5 control three-year-olds living in the catchment area but not selected to attend in the randomization. The three-year-olds in the treated areas are individually randomized.


58 control schools each with: five four-year-olds (not treated) from the catchment area and five non-treated three-year-olds.

Cohort 2: Same 60 treatment schools and 58 control schools as in Cohort 1. We recruit 5 three-year-olds in the catchment area of each treatment and each control school. The children belonged to families that were the closest five who consented to participate in the intervention

Minimum detectable effect size for main outcomes (accounting for sample design and clustering) Power 80%, False positive rate 5% Cohort 1: Comparison of treated and nontreated three-year-olds in treated school catchment areas (600 individually randomized 50/50) 0.23Sd. Comparison of four-year-olds in treated and non-treated school catchment areas (clustered, 600 Intracluster corr 0.2) 0.31Sd. Comparison of three-year-olds in treated and non-treated school catchment areas (clustered) 0.31 SD Cohort 2: Comparison of three-year-olds in treated and non-treated school catchment areas (clustered ICC 0.2) 0.31 SD

Pre-Analysis plan December 2025.pdf