Teaching at the Low Level: Experimental Evidence on Two Remedial Literacy Interventions in El Salvador

T1) 224 students from 28 schools and 202 households receive twice-weekly reading and writing sessions during regular school hours. Lessons are delivered by external, NGO-hired and trained teachers for 6 months. Each session has a duration of 90 minutes and is delivered in small groups of 10 to 15 students.

T2) 223 students from 28 schools and 202 households receive treatment 1 complemented by a parental intervention. Parents are motivated by twice-weekly text messages and focus meetings to support their children in their learning process.

The children in the control group (216 students from 202 households) take part in regular lessons and do not receive any specialized sessions.

2025-04-21

2025-10-21

Student learning outcomes in literacy

Student learning outcomes in literacy are measured via a baseline and an endline assessment that both consist of a written and an oral part. The assessments evaluate six core concepts: writing skills, motor skills, reading skills, reading comprehension, oral comprehension, phonological awareness.

Parental engagement
Parental responsiveness to intervention
Student motivation and engagement
Student grades in language, math and other subjects (if available)

We assess parental engagement by conducting a parent survey at endline.
We assess parental responsiveness to our intervention by tracking response rates to text messages and attendance at parent focus evenings.
We assess student motivation and engagement in our survey at baseline and endline and track student attendance of the specialized lessons.
We aim to assess whether we see any spillovers from the intervention to other subjects such as math by collecting student grades at baseline and endline. It is not yet clear, whether we will have access to that data however.

1. Baseline tests and a short socioeconomic survey are carried out with 2nd to 9th grade students who have difficulties with reading and writing, as evaluated by their teachers. The evaluation criteria used for baseline participation is whether a student has difficulties to either 1) read a short paragraph fluently and understanding what it is saying or 2) write a short sentence (including minor spelling mistakes).

2. The baseline is then evaluated and the 600 poorest performing students are included in the sample.

3. The children are then randomly assigned to one of the two treatment groups or the control group. In case various students from the same household participate in the study, we assign treatment at the household level to avoid substantial spillovers within a family.

4. For 5 months, the students in the treatment groups receive twice-weekly sessions from NGO-hired teachers during class hours. Parents of students from the second treatment arm receive a complementary intervention designed to benefit students.

5. Immediately after the treatment the childrens' reading and writing skills are assessed a second time in the endline test.

Randomization is conducted by the research team using the software R.

household (mostly one student per household)

No

606 households

663 students

T1) 202 households
T2) 202 households
Control) 202 households

We assume (conservatively) that the baseline covariates (e.g., the baseline test score, age, and gender) explain 50% of the variance. This calculation is based on data from two previous trials run by the research team in El Salvador, which found the explanatory power of age, gender, cohort and baseline test to be 0.45 and 0.69 respectively. Assuming 200 students per experimental group, we can therefore detect an effect size of at least (MDE) 0.198 SD with power 80% using a 5% test of level when comparing each treatment to the control or comparing the two treatments against each other.