Early Marriage and Perceived Risk of IPV

Adolescent girls and their parents will participate in a task. The task starts with a private and individual interview, followed by an information treatment (only for the treated households), and a second part of questions. Parents assigned to the treatment will be asked to watch a one-minute video that shows a husband physically abusing his wife in front of their children. The video will only be shown to the parents and not to the children of the households assigned to this treatment. The video is followed by the following information:
“A recent survey showed that in Bangladesh, about one in two women experience physical abuse by their husband. Girls who marry at a younger age are much more likely to experience abuse from their husband compared to girls who marry at an older age.”
After showing the video and reading out the statement, we will elicit the respondents' preferences in terms of the ideal age at marriage, and their beliefs about the likelihood of IPV with an increase in a bride's age and a bride's ability to resist IPV.
A few years later, we also plan to collect data on the actual date of the marriage of these adolescent girls. By randomly assigning households to the information treatment or a control, we will be able to identify its impact on the respondents' `preference' to delay the marriage, and possibly also the actual timing of the marriage. Comparing the perceived risk of IPV across a bride's age and the bride's perceived ability to resist it, we will examine possible mechanisms.

2026-01-04

2026-01-31

Adolescent girls' preferred age at marriage
Parents' preferred age at marriage for their daughters
Parent-daughter preference gap in age at marriage for daughter

Belief about the relationship between experiencing IPV and getting married at a younger age

We will run this experiment on different samples. First, around half of the households in our study were randomly assigned to BRAC's Social Empowerment and Legal Protection (SELP) programme, which supports girls' empowerment through after-school clubs. Assuming that SELP changes the girls’ aspirations, increases their awareness of the risks of child marriage and changes gender norms, we expect a difference in treatment effects, compared to the sample of households that were not exposed to this intervention. Second, in each household, we will randomly select the father or the mother for the survey experiment. This allows us to compare treatment effects between fathers and mothers. In sum, by randomly assigning households to the SELP programme or a control at the village level, next whether the IPV information with video is provided with village-level randomisation and finally randomly determining whether the father or the mother is interviewed (and always the daughter) with individual-level (i.e., adolescent-level) randomisation.

As a result, we will have eight groups: four groups for the control of the previous RCT and four groups for for the SELP arm of the previous RCT. We refer the groups for the control of the previous RCT as follows: TIFC (Information treatment+Father+control of the previous RCT), TIMC (Information treatment+Mother+control of the previous RCT),CIFC (Information control+Father+control of the previous RCT), CIMC (Information control+Mother+control of the previous RCT).

Similarly, we refer the groups for the SELP arm of the previous RCT as follows: TIFS (Information treatment+Father+SELP intervention), TIMS (Information treatment+Mother+SELP intervention),CIFC (Information control+Father+SELP intervention), CIMC (Information control+Mother+SELP intervention)

Not available

Randomisation done in office by a computer

Village level for information intervention assignment and individual level for parent assignment

Yes

250 villages (110 villages for the control from the previous RCT, 140 for SELP)

1800 in total

We will have four groups for each of control and SELP arms of the previous RCT: Information (mother), Information (father) Control (mother), Control (father). For the control arm of the previous RCT, each of these four group (referred to as TIFC, TIMC,CIFC, CIMC) will have 200 observations and for the SELP group, each group (referred to as TIFS, TIMS,CIFS, CIMS) will have 250 observations.