Can parental nudges improve early literacy in Sub-Saharan Africa? Experimental evidence from three contexts

Last registered on October 04, 2021

View Trial History

Pre-Trial

Trial Information

General Information

Title

Can parental nudges improve early literacy in Sub-Saharan Africa? Experimental evidence from three contexts

RCT ID

AEARCTR-0005569

Initial registration date

September 29, 2021

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published

October 04, 2021, 2:49 PM EDT

First published corresponds to when the trial was first made public on the Registry after being reviewed.

Locations

Country

Kenya

Region

Country

Nigeria

Region

Primary Investigator

Name

Daniel Rodriguez Segura

Affiliation

University of Virginia

Contact Primary Investigator

Other Primary Investigator(s)

PI Name

Benjamin Castleman

PI Affiliation

University of Virginia

Contact Investigator

PI Name

Beth Schueler

PI Affiliation

University of Virginia

Contact Investigator

Additional Trial Information

Status

On going

Start date

2021-07-01

End date

2022-02-28

Keywords

Education

Additional Keywords

JEL code(s)

I21 Analysis of Educational, I25 Education and Economic Development, I28 Government Policy

Secondary IDs

Prior work

This trial does not extend or rely on any prior RCTs.

Abstract

Education systems around the world have successfully leveraged additional family time and resources to further support children in the acquisition of their foundational literacy skills (e.g., Mayer et al, 2019; Teepe, et al, 2019 in the U.S., Knauer et al., 2019 in Kenya). However, these interventions tend to come with a relatively high price tag, and especially in low-income settings, can target parents with low educational attainment that may not be familiar with the best approaches to support their children’s education (Muralidharan et al., 2019; Portela and Atherton, 2020). In response to this, we evaluate an intervention that provides repeated low-cost reading materials, and low-touch behavioral encouragement to engage parents in their children’s literacy development process. We study the effectiveness of the repeated provision of properly leveled, low-cost, scaffolded materials (“postcards”) which guide parents through reading exercises they can perform with their children twice a week in an effort to improve early literacy outcomes. As such, we study this intervention through a randomized controlled trial of first graders during their first term across 112 private schools in Kenya, 42 private schools in Lagos (Nigeria), and 446 public schools in Lagos (Nigeria). Our experimental sample consists of over 23,000 students across these three settings.

Additionally, we investigate how the same educational intervention is implemented differently in three different settings. Contextual factors, implementation quality, and participant take-up are at the heart of the effectiveness of all development interventions. However, to date, much of the education literature in low- and middle-income countries has focused on the effectiveness, rather than on the rigorous documentation of the extent to which the local adoption and adaptation moderated their success. In this pre-registered analysis plan, we include a framework for how we will evaluate the implementation of the intervention in the three contexts based on our ex-ante theory of change. This framework will allow us to provide empirical evidence on the extent to which the same intervention was implemented differently in three contexts, and whether this moderated the effectiveness of the intervention.

Please see the attached document for further details on this study.

External Link(s)

Registration Citation

Citation

Castleman, Benjamin, Daniel Rodriguez Segura and Beth Schueler. 2021. "Can parental nudges improve early literacy in Sub-Saharan Africa? Experimental evidence from three contexts." AEA RCT Registry. October 04. https://doi.org/10.1257/rct.5569-1.0

Sponsors & Partners

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

Experimental Details

Interventions

Intervention(s)

The specific treatment consists of a “postcard” that students will receive every Monday throughout a full term (~15 weeks). This postcard contains two short reading passages, and one reading comprehension exercise on both sides and provides guidance for parents on how to engage on these passages and comprehension activities with their children. The content is largely the same across contexts, but the vocabulary and to a lesser degree the difficulty of the passages were adapted for each context. In an effort to maximize comparability across contexts, this intervention is set to start during the same school term (Term 1) across all three contexts.

The mechanics of the exercises are as follows: one side of the postcard is designed to be completed the same Monday the child brings the postcard home, and the other side is designed to be completed the following Thursday. Every Thursday afternoon, our partner will send parents an SMS text reminding them to complete the exercise. In an effort to increase perceived accountability of parents, each side of the postcard will have a designated space for parents to sign and acknowledge that they completed that day’s exercise. Furthermore, every Friday, students are encouraged to bring the signed postcards back to their teachers to check for completion.

Intervention Start Date

2021-07-01

Intervention End Date

2022-02-28

Primary Outcomes

Primary Outcomes (end points)

1. Oral reading fluency (measured in units of correct words per minute)
2. Standardized school grades for English and Kiswahili around the midterm (~6-7 weeks after the expected start of the intervention) and endterm (~13 weeks after the expected start)
3. Student dropout
4. Teacher survey on project implementation, and teacher beliefs and attitudes, also to be used to explore mechanisms
5. Parent survey on project take-up, and parent beliefs and attitudes, also to be used to explore mechanisms

Primary Outcomes (explanation)

The outcomes for oral reading fluency, language test scores, and enrollment come from our partner’s administrative data. The internal data collection process is mostly standardized within our partner’s network, allowing us to make direct comparisons for most outcomes and indicators across contexts. Specifically, while oral reading fluency and enrollment are coded in the same units across all three sites, the language test scores reflect different assessments by context, so we plan to use the standardized scores (by context) when pooling outcomes for all contexts. Similarly, within each context, all children take the same test within a testing round (i.e., during the midterm or endterm), allowing for comparisons across schools within each context.

We will also be studying the differences in implementation measures and take-up across contexts. Our main data source for this portion of the study is the teacher and parent phone surveys that will be carried out at the end of the intervention in each context. In the attached appendix, we pre-register a framework that outlines the expected theory of change of this intervention in 10 steps (from “1. Postcards are designed and adapted for the context” to “10. Students’ early literacy outcomes increase as a result of the postcard”). We also include in the appendix the survey instruments, and a clear mapping of each step along the theory of change to the survey questions we will use to analyze each step.

The survey data will also provide information on teacher and parental beliefs, which will allow us to explore factors that moderate and mediate the quality of implementation, and the effectiveness of the intervention. On the qualitative side, we will also have access to internal operations reports on the distribution of postcards to each school, as well as qualitative information from spot checks on the ground for a few schools. Pending travel restrictions, this set of qualitative data will be complemented with on-the-field interviews with teachers, principals, and parents.

Secondary Outcomes

Secondary Outcomes (end points)

To explore potential heterogeneity of our results, we will use individual level data on gender, school, and baseline grades. Similarly, we also use regional data on community size (population), poverty levels, adult female literacy, and relative wealth within each country.

Secondary Outcomes (explanation)

Beyond the specifications outlined above to directly answer our main research questions, we plan to conduct additional exploratory heterogeneity analyses with the covariates of interest within simple OLS models similar to the one described for the first research question. These covariates include local poverty, wealth, population, adult female literacy, baseline performance, gender of students, and age of students.

The vast majority of the data used for this project consists of administrative data provided by our partner. The data that our partner will provide consists of individual-level covariates for students in the treated and control groups which include the gender, and age of the student, school that they attend, time enrolled at schools run by our partner, baseline grades and grades from a previous year if available, and outcome variables. The school-level administrative data consists of the school latitude and longitude, the total enrollment, the pupil-teacher ratio in the overall student body and in the target grade, the female-male ratio in the overall student body and in the target grade, the average principal and teacher attendance rate, along with student attendance rates by gender. We complement the administrative data with geospatial data containing information on community-level covariates. In particular, we use the GIS poverty rate raster layer from Tatem et al. (2013), and the GIS adult (15-49) female literacy rate raster layer from Bosco et al. (2017). Both of these layers are at a resolution of 1-km at the equator. Similarly, we use Chi et al. (2021) to obtain georeferenced measures of local wealth. We use the latitude and longitude of each school to create an average poverty rate, average relative wealth, average adult female literacy rate, and total population for the 3-kilometer circular area surrounding each school.

Experimental Design

To minimize potential spillovers of the treatment, we randomly allocate treatment at the school level. To increase power, we stratify within each context based on the wealth level of the community surrounding each school, and the relative baseline performance. Within each strata, we randomly select one school to be treated.

Experimental Design Details

Randomization Method

Randomization performed using Stata 17 by researchers.

Randomization Unit

Schools

Was the treatment clustered?

Yes

Experiment Characteristics

Sample size: planned number of clusters

602 schools. Of these, 112 schools are in Kenya, 42 in Lagos (private), and 448 in Lagos (public). The appendix includes a table with descriptive statistics and sample balance across treatment and control schools for all three contexts. The appendix also provides visual representations of where these schools are geographically located.

Sample size: planned number of observations

Approximately 23,200 students across all three contexts.

Sample size (or number of clusters) by treatment arms

56 treatment and 56 control schools in Kenya. 21 treatment and 21 control schools in Lagos (private). 140 treatment schools and 308 control schools in Lagos (public). In total, there are 217 total treatment schools, covering approximately 8360 treated students.

Minimum detectable effect size for main outcomes (accounting for sample design and clustering)

Pooling students from all three contexts, we estimate that having on average 38.5 students per school across all 602 schools in the sample, with 36% assigned to treatment, assuming an R2 explained by individual and school-level covariates to be 0.30, and intra-class correlations of 0.20, consistent with our previous work with this our partner, we would expect a minimum-detectable effect (MDE) of 0.09 standard deviations (SD) for the pooled sample. The upper estimate of our power calculations is if we were to analyze data from each context separately. Using the same assumptions but replacing the average number of students per target grade and number of schools with the appropriate figures for each context, we would expect an MDE of 0.21 SD for Kenya and 0.35 SD for Lagos (private), and 0.11 in Lagos (public).

Supporting Documents and Materials

IRB