Are Parenting Interventions Transferable Across Settings? Evaluating Key Constraints in Sub-Saharan Africa

How to cite

Lichand, Guilherme; Wolf, Sharon. 2018. "Are Parenting Interventions Transferable Across Settings? Evaluating Key Constraints in Sub-Saharan Africa." AEA RCT Registry. AEARCTR-0003385. https://doi.org/10.1257/rct.3385-2.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Guilherme Lichand* , Sharon Wolf

* Primary investigator

Registry ID
AEARCTR-0003385
DOI
https://doi.org/10.1257/rct.3385-2.0
First registered
Last updated
Status
Completed
Trial dates
2018-10-08 – 2019-07-31
Countries
Côte d'Ivoire (Aboisso and Bouafle)
Registry record
http://www.socialscienceregistry.org/trials/3385

Abstract

While evidence from Brazil suggests that SMS messages to nudge parents’ engagement in their children’s education have large effects on educational outcomes, such an intervention might not work as intended in poorer settings, for at least two reasons: parents have a much higher likelihood of being illiterate – such that text-based interventions may fail to induce behavior change –, and teachers have less formal training and a much higher probability of being absent from schools – such that learning outcomes may not improve even if student’s attendance increases. This paper investigates whether those reasons are critical constraints for transferring the intervention across settings, in the context of Ivory Coast, by randomly assigning whether parents receive nudges over text or voice messages, and by cross-randomizing nudges across parents and teachers.

Intervention

The intervention, which has been designed and will be implemented by MGov and the Ministry of Education of Côte d’Ivoire, will randomly assign students’ parents and teachers to receive messages by the schools and by MGov. In particular, parents will be assigned to one of three treatment groups:

1. An “audio treatment”, in which parents will receive in the form of audio messages: (i) up to one message per week sent by school to parents about students’ attendance or performance, and (ii) two messages per week sent by MGov, as nudges, where parents will receive suggestions of simple activities that aid in social-emotional development, and which do not demand any curricular knowledge.

2. A text treatment in which the same information will be provided in the form of a message over SMS,

3. A control group, in which parents will not receive any message.

Additionally, teachers will also be assigned to one of two treatment groups:

1. A text treatment in which they will receive weekly messages over SMS with tips on activities to do with students and way to customize their classes to increase children’s learning,

2. A control condition, in which they will not receive any message from the platform.

Experimental design

The intervention will be evaluated through a school-randomized control trial with 100 schools in 2 regions in Côte d’Ivoire. In order to minimize spillovers, randomization will be done at the school level, which will be assigned to one of six treatment groups.

Cell 1: No communication with either parents or teachers
Cell 2: Communication with teachers but not with parents
Cell 3: Communication with parents (via text messages), but not with teachers
Cell 4: Communication with parents (via audio messages), but not with teachers
Cell 5: Communication with parents (via text messages) and with teachers
Cell 6: Communication with parents (via audio messages) and with teachers

Sample design is as follows:

Cell 1: 25 schools - 50 teachers + 5000 students (625 direct assessments)
Cell 2: 25 schools - 50 teachers + 5000 students (625 direct assessments)
Cell 3: 12 schools - 24 teachers + 2400 students (300 direct assessments)
Cell 4: 12 schools - 24 teachers + 2400 students (300 direct assessments)
Cell 5: 13 schools - 26 teachers + 2600 students (325 direct assessments)
Cell 6: 13 schools - 26 teachers + 2600 students (325 direct assessments)

We will use administrative data on test scores (if made available), attendance, and drop-out rates, as well as primary data collected through surveys with parents, teachers and head teachers, and direct assessments with students. Specifically, data collection will occur through: (i) direct child assessments administered in schools, with 25 randomly chosen students per school, (ii) surveys with teachers and head teachers administered in school, and (iii) directly administered surveys with caregivers in their home. This will occur at two time-points: baseline (October 2018 for parents, when assessors will already be going to communities for program enrolment, and October 2018 for children and teachers, during the first month of school), and follow up (June-July 2019, at the end of the school year). Finally, we will also administer a head-teacher survey at baseline, to assess school characteristics that may serve as barriers or supports for the successful implementation of the program.

Design details

Randomization method
Randomization done in office by a computer
Randomization unit
School-level randomization for treatment assignment Student-level randomization within schools for selecting subjects for direct assessments
Clusters
100 schools
Observations
20,000 students and 200 teachers
Arms
Cell 1: No communication with either parents or teachers; 25 schools - 50 teachers + 5000 students (625 direct assessments) Cell 2: Communication with teachers but not with parents; 25 schools - 50 teachers + 5000 students (625 direct assessments) Cell 3: Communication with parents (via text messages), but not with teachers; 12 schools - 24 teachers + 2400 students (300 direct assessments) Cell 4: Communication with parents (via audio messages), but not with teachers; 12 schools - 24 teachers + 2400 students (300 direct assessments) Cell 5: Communication with parents (via text messages) and with teachers; 13 schools - 26 teachers + 2600 students (325 direct assessments) Cell 6: Communication with parents (via audio messages) and with teachers; 13 schools - 26 teachers + 2600 students (325 direct assessments)
Primary outcomes
We will document the effects of the treatments on the following outcomes categories for students enrolled in the second and fourth year of primary school (aged ~7 and ~9 years old respectively), the final grades of the first two primary school cycles : 1. Students’ school attendance, grade retention and drop-out rates measured by administrative records and/or by schools’ inputs at MGov’s platform; 2. Students’ literacy and numeracy skills, measure through direct assessments; 3. Students’ cognitive performance in tasks aimed at assessing working memory and attention: • Child visual attention, measured through a stroop-task; • Child visual working memory, measured through a picture span task. 4. Students’ socio-emotional and self-regulatory skills: • Child self-reported social-emotional skills, directly assessed through items used in IDELA; • Child self-reported impulsivity scale. 5. Additional students’ outcomes: • Child engagement in labor activities, as reported by the child and by the parent, measured through the questionnaire. developed by Dr. Kaja Jasinska and has already been used in Côte d’Ivoire, • Child self-reported motivation, • Child self-reported self-esteem, • Child self-reported mindset (growth or fixed), • Child self-reported time use. 6. Parents’ outcomes: • Parent self-reported hypothetical willingness to pay to receive weekly messages about their child’s school life; • Parent’s involvement in their child’s education, as reported by the parent and the child, in terms of time spent for school-related activities • Parent self-reported mindset with respect to children (growth or fixed) and failure mindset, • Parent self-reported expectations, aspirations for their child’s education, and beliefs on the child’s school performance and attendance, • Parent self-reported mental health, • Parent self-reported discipline practices, 7. Teachers’ outcomes: • Teacher self-reported mindset and failure mindset; • Teacher self-reported motivation; • Teacher self-reported job satisfaction; • Teacher’s attendance (measure in three ways: (i) self-reported, (ii) reported by the students, and (iii) as reported by administrative records and/or by schools’ inputs at MGov’s platforms). Additionally, we will assess how students’, their parents and teachers‘ baseline characteristics moderate children’s response to the different versions of the program (audio and text), both in terms of outcomes measured by administrative records (grade retention and school attendance), but also their learning outcomes in both academic and behavioral domains. Since we have several outcome variables for each outcome categories, we will conduct a multiplicity of tests within each category. Estimating separate regressions for each outcome would substantially inflate the probability of false positives above stated significance levels. For this reason, we will build summary measures for each outcome category above, 1 through 7, Following Kling, Liebman and Katz (2007), we will normalize all outcomes to z-scores, and run seemingly unrelated regressions (SUR) to compute effect sizes for each summary measure, within outcome category.

Cited by

No citation data available for this registration yet.

Keywords