Child Marriage Law, Gender Norms and Marriage Customs
Last registered on June 14, 2018

Pre-Trial

Trial Information
General Information
Title
Child Marriage Law, Gender Norms and Marriage Customs
RCT ID
AEARCTR-0003035
Initial registration date
June 08, 2018
Last updated
June 14, 2018 5:56 PM EDT
Location(s)
Region
Primary Investigator
Affiliation
University of Kent
Other Primary Investigator(s)
PI Affiliation
University of Malaya
PI Affiliation
University of Kent
Additional Trial Information
Status
Completed
Start date
2018-05-27
End date
2018-06-10
Secondary IDs
Abstract
We propose to investigate the effects of child marriage law on marriage customs making use of (i) a new child marriage law in Bangladesh with separate progressive and regressive components, (ii) an existing nationally representative survey on women in Bangladesh with information on attitudes towards traditional marriage practices conducted immediately before the start of public debates that led to the new child marriage law, and (iii) a proposed exogenous treatment aimed at accelerating knowledge transmission about the new law in rural areas.
External Link(s)
Registration Citation
Citation
Amirapu, Amrit, Niaz Asadullah and Zaki Wahhaj. 2018. "Child Marriage Law, Gender Norms and Marriage Customs." AEA RCT Registry. June 14. https://www.socialscienceregistry.org/trials/3035/history/30756
Experimental Details
Interventions
Intervention(s)
There is research supporting the idea that the formal law can affect social norms via an "expressive effect" (e.g. Chen and Yeh, 2014). The main goal of this project is to investigate whether receiving information about a change in the formal law relating to child marriage can effect individuals' attitudes towards child marriage - in an environment characterized by low levels of enforcement. To this end, we randomly expose individuals to information regarding the elements of a child marriage law that was recently passed in Bangladesh, and which had both progressive and regressive elements. We then seek to test whether treated individuals have measurably different attitudes towards child marriage. Information about the law is transmitted via a short two minute video we produced for this purpose which dramatizes a hypothetical child marriage case while including information about either the progressive elements of the new law (Treatment 1), the progressive and regressive elements of the new law (Treatment 2), or only information about the old child marriage law (Control). We also randomly vary whether zero, one or two other members in the respondents' social network are similarly treated - in order to see whether peer effects may contribute to the formation of social attitudes.
Intervention Start Date
2018-05-27
Intervention End Date
2018-06-10
Primary Outcomes
Primary Outcomes (end points)
Our key outcomes are individuals' attitudes towards child marriage. Because it is challenging to measure attitudes about socially sensitive topics such as child marriage, we are attempting to do so in a variety of ways:
1) direct questions about respondents' attitudes regarding child marriage, including questions about:
o appropriate age of marriage
o whether a girl should have any say in choice of partner
o ideal age gap between a husband and a wife
o respondents’ beliefs, regarding his/her own community, about: prevalence of early marriage, parental expectations about daughters’ marriage timing, social attitudes towards young versus older brides
o attitudes towards traditional gender roles, using composite indices, as proposed by Asadullah and Wahhaj (2017)
2) questions about respondents' attitudes regarding hypothetical child marriage cases - as described by three vignettes. Questions cover respondents' own choice regarding hypothetical marriage decision; beliefs about what choices others would make; and approval or disapproval of a particular choice of others.
3) D-Score from an implicit association test (IAT) - designed for this study - to test attitudes towards child brides
4) decisions regarding how much to contribute towards a charity which works to help victims of child marriage and prevent child marriage
Primary Outcomes (explanation)
The only one of our outcomes that will be constructed is the implicit association test (IAT) score. For this we will focus on the "D score", which is a standard way of measuring outcomes from IATs (Nosek et al, 2007).
Secondary Outcomes
Secondary Outcomes (end points)
Our secondary outcome includes any steps taken to marry daughters (engagement, actively seeking a groom, discussions with family members about finding a groom) below the age of 18 in the six month period following the information intervention
Secondary Outcomes (explanation)
The marriage-related decisions include a range of outcomes including (i) a formal engagement, (ii) actively seeking a groom, (iii) having discussions within the family about finding a groom.
Experimental Design
Experimental Design
Our experimental design is as follows. We begin with a nationally representative sample of households with women aged 20-39 years in Bangladesh. From this sample, we randomly picked 24 districts (using Stata's random number generator). From these 24 districts, we then picked 80 villages randomly, but with greater weight on those villages with more marriage-age girls (again using Stata's random number generator). The study population consisted of all households from the original sample that were included in the 80 randomly selected villages. This amounts to approximately 920 households.
From this population, we randomly allocated one third of households to treatment one (T1), one third of households to treatment two (T2), and one third of households to a controld group. Each group was shown a video pertaining to child marriage containing different information about a new law. In addition, we implemented a second treatment orthogonal to the first: we randomly allocated households (again in population thirds) to receive a peer treatment: either 0, 1 or 2 of the respondents' peers would also be shown the video. Respondents were told how many of their peers would receive the treatment as well. Once the treatments are administered, respondents will be asked a series of questions, including questions to measure attitudes towards child marriage.
Experimental Design Details
Our experimental design is as follows. We begin with a nationally representative sample of households with women aged 20-39 years in Bangladesh. From this sample, we randomly picked 24 districts (using Stata's random number generator). From these 24 districts, we then picked 80 villages randomly, but with greater weight on those villages with more marriage-age girls (again using Stata's random number generator). The study population consisted of all households from the original sample that were included in the 80 randomly selected villages. This amounts to approximately 920 households. From this population, we randomly allocated one third of households to treatment one (T1), one third of households to treatment two (T2), and one third of households to a controld group. Each group was shown a video pertaining to child marriage containing different information about a new law. In addition, we implemented a second treatment orthogonal to the first: we randomly allocated households (again in population thirds) to receive a peer treatment: either 0, 1 or 2 of the respondents' peers would also be shown the video. Respondents were told how many of their peers would receive the treatment as well. Once the treatments are administered, respondents will be asked a series of questions, including questions to measure attitudes towards child marriage.
Randomization Method
Randomization done in office by a computer program. In particular, random numbers were generated across households in Stata, on the basis of which households were grouped into either treatment 1, treatment 2, or the control group.
Randomization Unit
Randomization was done at the level of the household.
Was the treatment clustered?
No
Experiment Characteristics
Sample size: planned number of clusters
80 village clusters in 24 districts in Bangladesh - excluding urban areas and the Chittagong Hill Tracts.
Sample size: planned number of observations
Approximately 960 initial respondent households, with approximately 960 additional respondents treated as part of the peer effect aspect of the study.
Sample size (or number of clusters) by treatment arms
Among the 960 initial respondents, 320 will be in treatment 1, 320 in treatment 2 and 320 in the control group. A similar breakdown applies to the 960 additional "peer effect" respondents. Numbers may not be exact due to the nature of sample selection: we first randomly sampled 24 districts, and then randomly sampled 80 villages within those districts - but with greater weight on villages with a large number of girls of "marriage-able age", in order to maximize the number of girls of marriageable-age included in the study.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
For a two-sample means test in which the control group has an outcome with mean 0 and standard deviation 1, and assuming 320 observations in the control group and 320 in the treatment, we will be able to detect a minimum effect size of .22 standard deviations (assuming alpha = .05 and power = .8).
IRB
INSTITUTIONAL REVIEW BOARDS (IRBs)
IRB Name
University of Kent Research Ethics Advisory Group
IRB Approval Date
2018-05-24
IRB Approval Number
0611718
Post-Trial
Post Trial Information
Study Withdrawal
Intervention
Is the intervention completed?
No
Is data collection complete?
Data Publication
Data Publication
Is public data available?
No
Program Files
Program Files
Reports and Papers
Preliminary Reports
Relevant Papers