Educational inequality and parental involvement during the COVID-19 pandemic: Randomised controlled experiment of a telementoring program in rural Bangladesh

In this tele-mentoring program, current university students from various educational institutions serve as a volunteer mentor for a maximum of two children over a 12-week period. They make a weekly 30-minute call at a pre-determined time of the day. As the children in this tele-mentoring program are primary graders, mentors mostly call, discuss and interact with the parents (mothers) to provide them with guidance and support. However, children accompany their parents during each mentoring session. The tele-discussion covers weekly goal setting (involvement time, study target), parenting assistance (solving textbook problems, idea for better psychological, emotional and social involvement, mental health support, etc.), and general counseling. We have provided all relevant textbooks and solutions (all digital versions) and training using online platforms to the mentors before the program begins. Beside the direct phone calls by the mentors, there will also be a text message campaign, which basically nudge the parents further to follow the mentor’s advices and guidelines.

2020-09-04

2020-12-03

There are two primary outcomes, one, children’s cognitive ability, and two, parental involvement in children’s educational and non-educational activities.

Children’s cognitive ability will be measured using standard assessment test based on the national curriculum of Bangladesh. The assessment test would consist of questions related to literacy (reading, vocabulary & speaking in Bangla and English) and numeracy. Due to the pandemic and the uncertainty regarding when it is safe to conduct face-to-face interview and assessment, the assessment will be conducted over-the-phone. The test totals 100 points which are divided equally between literacy and numeracy. The answers of the test will be in binary form to reduce the assessment bias.

Parental involvement in children’s educational and non-educational activities will be measured by interviewing the mothers. The interview will cover, two sets of questions. First set will cover the parental involvement time in hours with children in teaching writing and reading, storytelling, solving math problems, etc. The second set will cover the parental involvement time with children in playing, discussion, leisure time activities, etc.

There are two secondary outcomes, one, children's social preference, and two, parental perception and parenting style.

Children’s social preferences will be measured using standard lab-in-the-field economic experiments but administered over the phone. We use three experiments where children make decisions as a spectator so that the experiments can be executed over the phone. These three tasks are related to distributive justice. The decisions made in the distributive justice experiment will not affect the earnings of the participants but have real consequences on the gift received by other children. Three alternative conditions of distributive decisions are efficiency, luck, and empathy.

The parental perception will be measured in domains of negative parenting incidents (i.e. use of abusive words, scolding, beating, hopeless feeling about the child, etc.), Perception about parenting ability, Parental perception about their child’s future education, parenting style (authoritative, authoritarian or permissive), and confidence about parental involvement.

We use a parallel-group randomized controlled experiment to measure the effectiveness of this tele-mentoring intervention. To conduct this trial, we collaborate with a local research-focused NGO, Global Development and Research Initiative (GDRI). GDRI recently completed a randomized controlled experiment on early childhood education, named ‘investing in our future’ in southwestern Bangladesh, which was funded by UK ESRC-DFID research grants (AEA registry number: AEARCTR-0002356). More than 6000 children participated in that intervention in various experimental arms. In the first phase of this present tele-mentoring project, 1047 parents-child pairs from the previous’ project were randomly selected and interviewed. We then exclude families with non-eligible children, and parents who were unable or expressed unwillingness to attend the program, we end up with 840 parent-child pairs for the tele-mentoring program. Half of these 840 parent-child pairs are randomly assigned to receive the intervention, while the rest serves as the control group. Note that we already have assessment results of these children and survey data about the household’s socioeconomic status and parents’ involvement and interaction with their children from the previous project. Using the existing data, we find that characteristics of treatment and control groups are balanced.

Upon program completion, we will survey parents to understand their involvement in their children’s education and as well as their perception about parental engagement. Furthermore, after one month from the program’s end date, we will assess the children to measure their cognitive and non-cognitive outcomes and examine the effects of treatment on these outcome variables.

Randomization is done on an office computer using Stata’s random number generator.

Individual

No

840 parents-child pairs

840 parents-child pairs

420 parents-child pairs

We assume three alternative effect size of 0.20SD, 0.25SD and 0.30SD, 80 percent power, and a type-1 error rate of 5 percent. We base on an evidence from Botswana during COVID-19 where Angrist et al. (2020) have found 0.29SD effect of a similar program with shorter duration. To achieve this range of effect 176 to 394 observations are required in each arm. Accounting for some attrition from the program and sample, we include 420 children in each of the treatment and control groups.

Pre Analysis Plan