Human Capital at Home: Evidence from a Randomized Evaluation in the Philippines

The intervention, called "Family Academy" is a family coaching program to offer parents the knowledge and skills to prepare children for kindergarten and primary school. The first week of engagement between the FA coach, parent, and child consists of a pre-coaching assessment, after which there are eight weeks of math and phonics sessions. In each session, the coach engaged the parent in a four-step process. In the first half of each session, the coach taught a game or activity to the parent, often with the child watching or in the general vicinity. In the second half, the parent would teach the game or activity to the child. The coach would affirm the parent and child with positive feedback on the session. Finally, the parent was encouraged to offer positive feedback to the child during their session. The educational games included math cards, a number chart, and posters of colors and shapes for early mathematics and numeracy development, as well as cards and posters of the alphabet, parts of the body, family members, and foods in the local language to support early literacy and oral language development. The math cards were given to the household during the first session. In most cases, the math and phonics sessions were implemented as two separate sessions each week, so a household typically received 16 total visits (eight math and eight phonics) from the coach over eight weeks. Each session was intended to take 45 minutes. Following the final session, a
post-coaching assessment was conducted.

2021-09-01

2021-12-01

Learning outcomes

Four sub-tasks were used to measure preschool children’s early literacy and phonics skills. Eight subtasks were used to measure preschool children’s early math skills, adapted from Dulay et al.’s (2019) work with the Arcanys Early Learning Foundation. For phonics, we include alphabet knowledge by asking children to sing or recite the alphabet in either English or Tagalog, the main local language. Children were also tested on their ability to identify letters of the alphabet across 20 letters. A key measure of interest was the child’s ability to identify beginning sounds of words. To identify this measure, the child was shown a picture of an object, followed by the assessor speaking the word out loud and asking the child to identify the beginning sound. Another sub-task required children to point to the letter that corresponds to a given sound.

To assess math skills, participants were asked to visually identify as many colors as possible among ten colors. For shape identification, children were asked to similarly identify eight different shapes. To examine children’s understanding of counting and early skills in cardinality, participants were asked to count to ten in either English or Tagalog. To measure children’s ability to object count, participants were asked to count the number of animals on different cards. Ten such subitems were assessed in total. To assess the ability to identify numbers, children were presented with a grid with numbers 1-10 in random order, which they were then asked to identify. To
understand the child’s ability to compare numbers, the child was presented with five sets of two numbers to compare. In each of these sets, the child was asked to identify whether numbers were the same, or which one was greater or lower in magnitude. To examine children’s understanding of numerical sequencing and patterns, children were given cards with a missing number in a sequence
of consecutive numbers, which the children were then required to identify. The final measure was designed to assess the child’s understanding of simple addition. Ten such items were assessed in total for this sub-task.

Parent-child interactions, parent beliefs, parent investments in education, parent labor market outcomes

The study was conducted across six regions in the Philippines: Bacolod, General Santos, Iloilo, Kalibo, Koronadal and Palawan shown in Figure A.1. 188 communities had at least four households with a 3-5-year-old child (for a total of 1609 households). We randomized at the community level: 91 treatment communities (788 households) and 97 control communities (821 households).

Randomization done by an office computer

Community

Yes

188

1609 households

The study was conducted across six regions in the Philippines: Bacolod, General Santos, Iloilo, Kalibo, Koronadal and Palawan shown in Figure A.1. 188 communities had at least four households with a 3-5-year-old child (for a total of 1609 households). We randomized at the community level: 91 treatment communities (788 households) and 97 control communities (821 households).