Improving early literacy outcomes in Guatemala

This study comes from the randomized implementation of the “Spark Program” by Cooperativa para la Educación – Guatemala (“CoEduc”). The Spark Program has been in place for over 10 years, and has been rolled out to schools throughout different municipalities in Guatemala as funding becomes available to CoEduc. Specifically, Spark focuses on developing children’s early literacy skills, given the dire state of literacy in rural Guatemala. Spark provides schools with a set of high-quality, culturally-sensitive, and curriculum-aligned textbooks and materials, tailored to the Guatemalan curriculum for grades 1 and 2. The books come with intensive teacher training and in-class coaching on the best strategies to utilize these materials. Each teacher receives three trainings per year, spaced out throughout the year. The teacher trainings and coaching sessions are mandatory for all teachers teaching a focal grade (1-2) in a school that has the Spark program. Overall, the cost of the program is about $90 per student per year.

2018-01-01

2019-11-30

A local adaptation of the EGRA test was used to measure early literacy skills at baseline and endline of both 2018 and 2019.

Oral language, alphabetic principles, decoding, phonological awareness, rapid automized naming, reading fluency, reading comprehension, writing, and an indexed outcome for all of these

Attitudes towards literacy

Whether student likes to read and write, whether student can finish

The precise setting for this program was CoEduc’s planned expansion to the Tecpán municipality in 2018, where they allowed public schools to express interest in joining the program. After schools signed up, the number of interested schools exceeded their logistical capacity for yearly rollout. Specifically, 15 schools expressed interest in implementing the Spark program, but CoEduc only had capacity for about half of the schools. Therefore, entry into the program in 2018 was randomly allocated, and 7 schools were selected for expansion into Spark in 2018. In 2019, the rest of the 8 schools entered the program. Therefore, the sample consists of 15 schools, all of which entered the Spark program in either 2018 or 2019. Randomization was performed by CoEduc.
Among the schools that entered the experimental sample, CoEduc randomly sampled at least 12 students per focus grade (first and second grade), which led to an average of ~28 sampled students per school across both grades, for a total of ~28 students across all 15 schools for 2018. For 2019, the sample consists of ~450 students, approximately half in each grade. All students are sampled at baseline, and followed to the end of the year. The students in second grade in 2018, exited the program in 2019 and therefore a follow up was not possible. The students in first grade in 2018, were tracked to the best of CoEduc’s abilities into 2019 for baseline and endline measurement. Approximately 4 in 5 of the first graders in 2018 were surveyed again in 2019, and the rest of the 230 students making up the sample in second grade in 2019 were randomly selected to fill in the place of the students that CoEduc was not able to locate. In 2019, there was also a new cohort of first graders into the Spark program, although by 2019 Spark had expanded to all schools in the sample, leaving no pure control group in 2019.

Randomization performed by NGO

School

Yes

15 schools

~400 students at baseline in 2018 with data, but all schools in treatment schools were treated.

~200 students in the treatment arm in 2018

With a full sample of approximately 420 children, 50% of them in treatment schools, across 15 schools, or ~28 students per school. Assuming R2 absorbed by covariates at level 1 and level 2. The MDE is 0.20 SD.