Parental Involvement in Education and Children's Human Capital Formation

The intervention is a school-based parental volunteering program implemented across 80 primary schools in Bangladesh. Schools sent invitation letters to about 5,000 parents describing an opportunity to volunteer at their child’s school for 2–3 hours whenever help was needed or during school events. Parents who agreed returned consent letters, and from these volunteers, a study sample was drawn. In 60 “intervention” schools, consenting parents were randomly assigned either to: (1) a treatment group that was formally signed up to volunteer in school activities for about 18 months (May 2023 to Sep 2024); e.g., helping organise sports day/cultural events, assisting with school programs, cleaning and other support tasks, and receiving weekly SMS reminders encouraging volunteering; or (2) an active-control group from the same schools who were told the volunteering slots were full and were not needed, receiving no SMS reminders. In 20 additional schools, parents followed the same recruitment process but were assigned to a pure control group with no volunteering role either. At the time of pre-registration, we are still not done with collecting data from schools and teachers.

2023-05-17

2024-09-26

Outcomes of children: Primary outcomes include---
(A) Cognitive Outcomes: (1) standardized test scores in mathematics and Bengali; (2) working memory; and
(B) Non-cognitive outcomes: (3) school attendance; (4) academic motivation, (5) happiness at school, (6) self-confidence about education, (7) socio-emotional functioning using the SDQ (excluding prosociality).

Outcomes of parents and potential channels for children’s outcomes: Primary outcomes include---
A. Parental Investments on Children: (1) family care and support for learning (FCI), (2) financial and time investments in children’s learning, (3) other time-intensive parental investments, (4) parental carefulness, (5) setting limits, (6) alternate care arrangements, (7) knowledge about children’s school, (8) beliefs about child’s performance relative to actual performance;
B. Parental Well-being: (9) mental health (stress and anxiety), (10) mothers’ bargaining power, (11) subjective well-being (happiness and life satisfaction), and
C. Parental Networks: (12) communication with other parents and social networks—which captures both the extensive margin (number of parents known, considered friends, or contacted for advice or support) and the intensive margin (frequency of contact, sharing of parenting or education tips, and discussions of personal or financial concerns)—(13) networks with other parents and (14) altruism (hypothetical dictator game)

In addition to this, we will also aim to measure teacher and school outcomes from as many schools as possible, depending on permissions and logistical constraints.

Outcomes of teachers: Primary outcomes include---
A. Teacher well-being (1) mental health, (2) subjective well-being, (3) absenteeism, and
B. Teacher investments (4) extra tutoring, (5) contact with parents (including recall of parents’ names as an indicator of relationship depth).


Outcomes of schools (by surveying principals and observations): Primary outcomes include----
(1) recent investments in schools and of what type, (2) teacher–student ratio per classroom (i.e., whether new teachers were hired), and (3) school or classroom quality checklist, including whether a teacher is present in class, whether the classroom is noisy, and whether teachers use corporal punishment.

All indices will be standardized relative to the active control group and constructed following Anderson (2008), with higher values indicating better outcomes. Indices will be standardized relative to the pure control group whenever regressions include this group.

Children: Secondary outcomes include---
(A) School Environment: (8) experiences of bullying or discouragement for studying, (9) prosociality measured through a Dictator Game.
(B) Academic Effort and Motivation: (10) school dropout, (11) academic self-perceptions, (12) self-study engagement, (13) proportion of hours allocated to study using time allocation across school, home study, play, and interactions with teachers or parents.

Parents: Secondary outcomes include---
A. Experience with Schools: (15) volunteering at schools, (16) communication with schools, (17) beliefs about teacher effectiveness
B. Parenting Practices and Aspirations: (18) child–parent relationship, (19) parenting style, (20) home study time, (21) satisfaction with children’s school performance, (22) aspirations for children’s education.

In addition to this, we will also aim to measure teacher and school outcomes from as many schools as possible, depending on permissions and logistical constraints.

Teachers; Secondary outcomes include---
A. Beliefs about Students: (5) aspirations about students’ education and (6) beliefs about student performance. In addition, we will also measure (7) perceptions of the volunteering program (including qualitative views on its impact and suggestions for improvement).

Schools: Secondary outcomes include---
(4) parents’ feedback and satisfaction, including any complaints about school quality or facilities, (5) whether schools are happy about parents’ volunteering and why (qualitative), (6) how they would like to improve or change the volunteering program (qualitative), (7) whether volunteering increased schools' expenditures and how (qualitative), and (8) which policies or changes are perceived as most effective for improving student performance and well-being, and where parent volunteering ranks among them (qualitative question).

All indices will be standardized relative to the active control group and constructed following Anderson (2008), with higher values indicating better outcomes. Indices will be standardized relative to the pure control group whenever regressions include this group.

We partnered with 80 public primary schools (grades 1–5) in the Rajshahi region of Bangladesh. About 5,000 parents were invited to volunteer at their child's school; roughly 70% consented and were eligible for sampling. From consenting parents, we randomly selected about 1,500 parent–child dyads for surveys.

Parent-level randomization within intervention schools: In each of the 60 intervention schools, around 22 consenting parents were randomly chosen and then individually randomized into:
- Treatment group (volunteering encouragement): 14 parents per school were formally signed up to volunteer for 18 months and received weekly SMS reminders encouraging participation.
- Active control group (no volunteering role): 8 parents per school were told volunteering slots were full and were not asked to volunteer; they received no SMS reminders.

Pure control schools: In each of the 20 pure control schools, 8 consenting parents were randomly selected into the study sample; no parent was asked to volunteer, and no reminders were sent.

Total schools N=80:
- 60 schools were "intervention schools".
- 20 schools were "pure control schools".

Planned sample sizes (pre-attrition): 860 treatment dyads, 481 active control dyads, and 160 pure control dyads.

Not available

Randomization was done in the office by computer

Individual level

No

80 schools

- 1501 parent-child dyads - Up to 160 teachers (up to 2 teachers per school) - Up to 80 headmasters (one per school)

-Treatment parents: 860 parent-child dyads
-Active control: 481 parent-child dyads
-Pure control: 160 parent-child dyads

We focus on the ITT comparison between treatment and active-control dyads within the set of intervention schools (i.e., individual-level randomisation). We consider children's cognitive outcome index (math, Bengali, and working memory) as the main outcome for this power calculation. Cognitive index will be in standardized units, so MDEs will be expressed in SD terms. Using a two-sided test with alpha=0.05 and 80% power, and assuming no baseline adjustment and an allocation ratio of 1.8 (i.e., N_t/N_c), detecting an MDE of 0.16 SD requires N_t=858 treated dyads and N_c=479 active-control dyads. Our sample of N_t=860 and N_c=481, therefore, meets this requirement. The MDE is motivated by evidence from the second-order meta-analysis by Kim (2022), “Fifty years of parental involvement and achievement research: A second-order meta-analysis” (in Educational Research Review, 37, 100463), which reports average impacts of about 0.16 SD from parental-involvement interventions targeting grades K–12. For comparison, effect sizes are larger in early-childhood parenting programs: Jeong et al. (2021), “Parenting interventions to promote early child development in the first three years of life: A global systematic review and meta-analysis” (PLOS Medicine, 18(5), e1003602), find an average effect of roughly 0.32 SD on cognitive development. Using MDE=0.32 SD, we would need N_t=217 treated dyads and N_c=121 active-control dyads. We calculated our samples assuming there are no positive spillovers on the active control dyads from the treatment dyads.