This study evaluates whether an intervention, namely Tipping Point (TP), can reduce child marriage and increase decision making capability among the adolescent girls in rural areas. This is a cluster randomized controlled trial with three intervention arms - full TP package, light TP package (differs from the full TP only in terms of the intensity of the social norms change component) and control. The study is being conducted in Pirgacha sub district of Rangpur district in Bangladesh. A 18-month intervention will be in place after completion of the baseline survey and the endline survey will be conducted 18 months after the intervention completion. This design will allow the investigators to measure the impact of full TP intervention, light TP intervention and the emphasized social norms change.

Background Worldwide, every year, an estimated 14 million girls aged under 18 are married with little or no say in the matter. Before age 15, girls are not physically or cognitively ready to make safe and consensual decisions about marriage, sexual relations, or reproduction. In middle adolescence (15 to 17 years), physical and cognitive readiness still varies widely, depending on the onset and pace of puberty, cognitive maturation, and the risks and responsibilities encountered at the time of marriage and childbearing . Marrying while physically and cognitively immature may limit a girl's educational opportunities, her social and emotional development, and her health and that of her children. Women who marry in childhood more often are underweight, report unintended pregnancy, experience obstetric complications and maternal death, have preterm or low birth weight births, and exhibit postnatal depression . These girls are more vulnerable to spousal physical violence. It is imperative to ensure human rights, development, health and well-being not only for the girls', but also for generations to come by addressing child marriage.

In South Asia, child marriage rates vary considerably. The rate of child marriage is highest in Bangladesh followed by Nepal. The percentage of girls marrying before age 18 and before age 15 in Bangladesh were 598.6 and 22.4 respectively and in Nepal the percentages were 4039.5 and 7.0 respectively . As elsewhere, the drivers include deeply entrenched cultural and religious beliefs, worries about family reputation and the opinion of others, dowry, poverty, parents' desires to secure economic well-being for their daughters and a perceived need to protect girls from harm, like sexual harassment and abuse. Pervasive fears of natural disasters also deepen communities' poverty and render them more vulnerable to practicing child marriage .

Preventing child marriage has been a global agenda and one of the major goals of Millennium Development Goals (MDGs) and Sustainable Development goals (SDGs). In line with MDGs and SDGs Governments, Non-government organizations (NGOs) and research organizations in different countries have attempted to tackle child marriage through different programs. Most of the interventions included girls' education, livelihoods/conditional cash transfer, empowerment, and community mobilization . Unfortunately, the majority of these interventions have not been evaluated rigorously . This results in missed opportunity to learn what actually works in reducing child marriage and what does not.

Although the body of evidence on how to prevent child marriage is growing, its rate did not decline at a desirable pace. Previous interventions targeted many of the root causes of child marriage. However, despite wide recognition of the need to change pro-child marriage social norms attempt to study the impact of social norms change in addressing child marriage was rarely investigated or investigated well. Lack of understanding of social norms and how to change them effectively impedes development of effective and sustainable child marriage prevention programs. Moreover, for most of the interventions, there was a not a rigorous evaluation . Again, the interventions that had strong evaluation design did not necessarily focus on social norms change .

At this backdrop the current research proposes to conduct a cluster randomized controlled trial (CRCT) to evaluate Tipping Point (TP), an intervention to address child marriage in Bangladesh using intervention packages that differ only in terms of the intensity of the social norms change component.

Objectives of the study While the main objective of this study is to measure the impact of TP intervention, the specific primary objectives are -

1. To assess the impact of full package of TP intervention (including emphasized social norms change) on reducing child marriage among the adolescent girls participating in TP and within community; and increasing decision making capability of the adolescent girls' participating in TP intervention.
2. To assess the impact of the light package of TP intervention (without emphasized social norms change) on reducing child marriage among the adolescent girls participating in TP and within community; and increasing decision making capability of the adolescent girls' participating in TP intervention.
3. To assess the impact of the emphasized social norms change on child marriage. The secondary objectives of this study are to assess whether both full and light TP interventions -

1. Increase adolescent girls' critical awareness regarding gender and rights
2. Promote adolescent girls' positive aspiration and progressive choices
3. Improve adolescent girls' communication and negotiation skills
4. Increase adolescent girls' knowledge, attitude and practice regarding SRHR
5. Bring positive change in social norms in terms of gender, rights and child marriage
6. Improve group cohesion, solidarity and mobilization skills among adolescent girls The Tipping Point Intervention To the best of our knowledge, TP is one of the very few attempts to develop a comprehensive evidence-based and theoretically driven intervention for preventing child marriage. A Community Participatory Analysis (CPA) conducted in the first Phase of the project found that in these communities, child marriage is rooted in a complex dynamic of limited choices and material resources (economic insecurity, lack of livelihood options, household labor) and prevailing social norms about family honor, control of sexuality, and the low social status of girls. Based on the findings of the CPA , Prevention of Child Marriage in Bangladesh is rested on two major approaches, the sexual rights approach and the social norms approach. A Rights based approach explicitly defines the sexual and reproductive rights in the overarching women's rights framework. CARE's Social Norms Approach focuses on not just generating awareness but also breaking from the psycho-social barriers created by normative and empirical expectations. These approaches frame the outcomes and the intervention strategies to be used.

The TP Theory of Change (TOC) focuses on four key areas for strategic intervention in order to reduce the incidence of child marriage:

1. Awareness generation

   o Adolescent Girls experience increased power over major life decision areas (marriage, education, engagement in livelihoods activities, bodies/sexuality)

   • girls' increased knowledge about equality, rights, SRHR, livelihood options, social norms;
   • girls' increased communication and negotiation skills

2. Developing and sustaining open channels of communication and discussion o Increased dialogue skills by all

   o Parents' increased knowledge of adolescents' aspirations

   o Fathers' increased awareness of how gender inequalities affect daughters, wives, sons

   o Mothers' increased skills to advocate for daughters

3. Breaking Gender Norms o Weakening of gendered social norms that limit adolescent girls' lives, particularly norms regarding girls' playing sports, interacting with boys, moving across and outside of their village, and individually and collectively asserting their rights to equality and to take major life decisions

   o girls, boys, mothers, fathers have increased awareness of the social norms affecting girls' lives and increased critical thinking about gender equality aspects of social norms

   o boys and men are aware of the role that hegemonic masculinities play in limiting their lives.

4. Developing Girl's Leadership capacity - Shift in dynamics of Power o Girls and parents' and other activist groups have increased skills for planning, implementing and evaluating collective action

   o Girls' increased cohesion and solidarity within the Girls' Groups

   • Girls in Activist training have increased leadership and mobilization skills

The overall Tipping Point intervention package will span 18 months and will involve:

1. Sessions with girls' and boys' groups: 60 weekly sessions, each of 90 minutes. Themes (and topics) will cover Gender and life skills (Mobility, knowing our communities, equity, gender roles and stereotypes, setting life goals, negotiation); ASRHR (Bodies, puberty, consent, pleasure and pain, healthy boundaries, family planning, technology and consent, gender based violence (GBV), life goals, etc.; Alternative livelihoods for girls (Financial literacy, discussions of non-stereotype income-generation opportunities, life goals, budgeting, saving, gendered division of labor); Girl-centred movement-building (Leadership, planning the social norms public activities, rights-based information, equity, mobility); and Social norms (Understanding social norms, planning for social norms change, menstruation norms, dowry, girls' voices and choices and aspirations, family honor.).

2. Facilitated sessions with mothers' and fathers' group: 18 monthly sessions, each of 90 minutes

3. Dialogue sessions jointly with girls' groups and fathers' groups, and girls' groups and boys' groups: 6 sessions. Typical example includes i) Dialogue between girls and fathers about girls' dreams and aspirations for their lives (as many fathers had never thought of their daughters' hopes for her future) and ii) Dialogue between girls' groups and boys' groups about boys' sexual harassment of girls and its impact on girls or reasons for boys why they may engage in this behavior.

4. Facilitated sessions with girls' activist group: 12 monthly sessions, starting in month 7. The theme (and topics) include trainings on running a campaign on a selected theme (choosing tactics and actions, and planning and budgeting for events). Village savings and loans associations (VSLAs) will be started and enrolment will be open to members of the girls' groups. VSLA includes training on basic financial literacy, promotes leadership skill and provides a low-risk opportunity for very poor households.

5. Facilitated sessions with boys'/parents' action group: 12 monthly sessions, starting in month 7. The theme (and topics) will include Knowledge and skills for activism (Leadership, community organizing principles, information on rights and citizen involvement).

6. Intensive facilitated workshop with selected religious leaders/influential people: 6 quarterly, each of 2 days. The topics will cover facilitated discussion and activities, with some individual follow up by frontline workers.

7. Community and Social Norms Change Component: there will be six "planning for a public social norms change activity" sessions with the girls' groups. In line with CARE's "Social Norms Design Principles" the Girls and Boys Groups will be motivated to design public activities and dialogues that challenge gendered social norms (e.g. girls' soccer and boat racing competitions, public dramas about girls' mobility or girls' speaking out for equal rights, boys cooking competition, fathers braiding their daughters' hair competition). Ultimately the girls' groups, with support from boys' and mothers' and fathers' groups, will organize at least six public social norms change activities of their choosing.

   The sessions/ activities/ manuals are synchronized. Most of the sessions will run parallel across groups- e.g. when the girls' groups are discussing gender-based violence, consent, sexual harassment, the boys' groups will also be discussing these topics from a masculinities/boys' perspective. Where possible, this will be followed by a cross-group dialogue session or planning session for a public activity.

   Tipping Point overall takes a SAA (Social Analysis and Action) approach , which includes an extensive Gender, Equity and Diversity (GED) training for staff as a basis for all the work. GED trainings are trainings in which staff reflect upon their own lives and their relationships with gender, power, diversity and equity. As the second step of the SAA approach, the staff will be fostering reflective analysis with the groups in the communities on the various topics, encouraging the groups to identify situations which they would like to change in their communities, then supporting them to plan actions related to specific topics, and later to evaluate their actions.

   As a minimum standard for facilitation, facilitators must have passed through GED and SAA training, as well as on trainings for facilitating the specific sessions. Boys' sessions will be facilitated by male facilitators and girls' sessions by female facilitators; parents' sessions will have at least two co-facilitators.

   Research Design and Methods

   This study will be implemented in one upazila (sub-district) at Rangpur district of Bangladesh. Rangpur has been selected as the median age at marriage is 15 years among women aged 20-49, which is the lowest in the country. It employs a three arm Cluster Randomized Controlled Trial (CRCT) design and uses both quantitative and qualitative methods for evaluating TP. The arms are as follows:

   Arm 1: Full TP intervention including emphasized social norms change Arm 2: Light TP intervention without emphasized social norms change Arm 3: Pure control

   This 3-arm design will enable the investigators to measure the effect of the light TP intervention as well as the effect of social norms change on TP outcomes. The effects will be determined by comparing the three arms as follows:

   Arm 1 - Arm 3 = Effect of full TP intervention Arm 2 - Arm 3 = Effect of light TP intervention Arm 1 - Arm 2 = Effect of emphasized social norms change The villages from the selected upazila will be treated as clusters or the primary sampling units in this study. The first cluster will be selected randomly from the list of villages under the upazila. The villages sharing borders with the first selected village will be considered as buffers. The next selected village will be the one located in the north-eastern direction from the first selected village dropping the buffer villages. This will continue until the required number of clusters has been chosen from the upazila. Then each cluster will be randomly assigned to one of the three arms.

   Qualitative study The qualitative study consisting of both baseline and endline will be carried out in four villages (two villages in each arm; village A and B in Arm 1 and village C and D in Arm 2) in two intervention arms of TP. Data in each arm will be collected using 4 Key Informant Interviews (KII) (2 men and 2 women); 10 In-depth Interviews (IDI) with adolescent girls and 5 IDIs with adolescent boys; 2 Focus Group Discussions (FGD) with adolescent girls, boys, adult women and men each. Use of 3 different techniques of data collection will allow data triangulation. While KIIs and FGDs will provide data on social norms and general practices IDIs will give access to in-depth data on individual perceptions, practices and experiences.

   The qualitative data collectors will receive a 7-day training on gender, rights, empowerment, child marriage, qualitative research methods, and research ethics. Interviewers will be gender matched with the informants. The endline will follow the same strategy enabling the investigators to assess changes over time in the cohort of group members and in the community over time in each of the arms.

   Quantitative evaluation Villages will be treated as clusters or the primary sampling units in this study. Each cluster will have one adolescent girls' group consisting of 22 girls. The rates of child marriage and the decision making ability among the target group (girls aged 12 to below 16) are unknown. So, the investigators assumed the prevalence rates as 50% for each outcome among this group. A life skills intervention in Maharashtra, achieved a 19% reduction in child marriage in the community . Based on this literature 15% effect size was considered along with 5% level of significance and 80% power. As it is a cluster randomized controlled trial, intra-cluster correlation (ICC) needs to be taken into account while calculating the sample size. The ICC differs for different outcomes. ICC for child marriage was found to be 0.05 using data from a icddr,b conducted nationally representative study. As information on ICC for decision making is unavailable, ICC value for child marriage was used for calculating the sample size. Considering 5% significance level and 80% power, the required number of clusters was 17 per arm making the total number of clusters 51. Considering a 15% non-response/dropout rate the group size will be increased to 25 and total sample size will be 1,275 girls.

   For the community survey for assessing social norms change, the assumptions included 50% prevalence of child marriage related norms (as it is unknown for the target group), 15% effect size 5% significance level, 80% power and 5% non-response rate. Thus, the required sample size was 90 adult females and 90 adult males aged 25 and above in each arm making the total sample size 540.

   Once the clusters are formed, enumeration of the households in each cluster will be conducted for collecting information on the household members' sex, age, marital status, age at marriage, and occupation. A properly trained team of data collectors will conduct this enumeration. Special care will be taken to train them on how to collect and check information on age and age at marriage. The enumeration data will provide the investigators with an estimation of child marriage in each cluster and with a list of eligible adolescent girls for group formation. The girls aged between 12 and under 16 years and currently living in the study village will be considered as eligible. The group members will be selected randomly from the list of eligible girls. All the group member adolescent girls will be included in the survey. These enumeration data will also serve as the sampling frame for the baseline and endline community surveys. Adult male and female community members who are currently living in the study village will be considered as eligible. Two separate surveys will be carried out: (1) girls survey among selected girls to collect information on all study outcomes; and (2) community survey among adult male and female community members to capture social norms around child marriage. The baseline survey of the adolescent girls will start once group formation is completed in each cluster. This cohort will also be interviewed at endline. For the social norms surveys in the communities, random samples of eligible adults from community will be selected and interviewed at baseline and endline. The surveys will use face-to-face interviews conducted by gender-matched interviewers. The survey team will receive a 12-day participatory training on gender, violence against women and girls, ethics, survey methods, the questionnaire, and use of tablets.

   The interviews will be conducted in private in a location convenient for the respondents upon their assent and consent of the parents in case of minors. The data will be collected using Personalized Digital Assistants (PDAs). Offline-based survey software will be developed. One programmer will be available during the survey period for necessary trouble shooting and data uploading. The data collected will be uploaded on a designated server at the end of the business day, which would allow the researchers to instantly review the data from the internet. Confidentiality will be maintained by keeping the identification information in separate files. Data will be password protected to avoid unauthorised access to data in case of unavoidable circumstances and will be used only for the purpose of research. All data will be anonymous during the time of analysis.

   Monitoring of TP Monitoring qualitative aspects of the intervention is of utmost importance. Data collection for qualitative monitoring may occur every 4 months during the intervention. Techniques for monitoring will include observation of group sessions of the adolescent girls', boys and parents (12 in each round); observations of public events (2 events, twice in whole intervention period); FGDs with the group session facilitators (4 in each round) and short interviews with the group members (8 in each round). Instant feedback to the facilitators after observation will help the facilitator in understanding their weaknesses and strengths and create an opportunity to improve performance. Observation reports will also be shared at higher levels making these levels aware of the weaknesses and strengths of the facilitators and the challenges they face. This will enable timely support to the facilitators. FGDs with the facilitators will help in identifying the challenges they face and in coming up with solutions collectively. Short interviews with the group members will enable the investigators to understand the challenges the group members face in attending, understanding and participating in the sessions.

   Quantitative data analysis Descriptive analyses will be performed to report frequencies and percentages of different outcomes of interest. Chi-square, ANOVA and t-tests will be performed to check whether the study arms are covariate balanced. The impact will be assessed using risk ratios derived from binary regression analyses for measuring change in outcomes in the intervention arms relative to change in the control arm. Mixed effects models will be constructed with cluster as random effects and time point, intervention group, and time point x intervention interaction as fixed effects. Adjusting for the dependent variables at baseline survey, it will provide a measure of the relative change in intervention group that took place between baseline and endline compared to the change in the control group.

   Qualitative data analysis With the informants' permission, all interviews and FGDs will be digitally recorded. All recorded interviews will be transcribed verbatim. Data analysis will be iterative. To explore research questions, the intense will be combined, within-case focus of Narrative Analysis with the across-case approach of Grounded Theory, a robust analytic strategy for qualitative data.

   Narrative analysis will be used with the IDIs to examine each transcript for a core narrative or 'story' of each adolescent girl and boys about how they experience and view gender; restrictions and barriers in fulfilling their aspirations; privileges and discrimination; agency and/or the lack of it; social norms and practices around child and forced marriage, drivers and triggers of child and forced marriage, etc. Core narratives across participants will then be compared to identify distinct and shared features of the narratives. Narrative analysis will identify, for example, the baseline status of the adolescent girls and boys in the mentioned areas. A comparison of the IDIs of the same girls and boys at baseline and endline will provide with insights into change that relates to TP. Grounded Theory (GT) is well suited 1) to identify cross-cutting themes and normative constructs related to our areas of investigation and 2) to link these themes into an explanatory framework to understand more fully the context of child and forced marriage.

   Recommended steps for GT analysis will be followed: 1) Identify themes and develop a codebook. The investigators will systematically read, memo, and discuss themes raised in the IDIs, KIIs and FGDs. Core themes will be developed into a codebook, listing each theme and how to label data for each theme. The codebook will include inductive themes emerging from the data and deductive themes developed a priori from theoretical domains in the interview guides. 2) Code data. Using themes from the codebook, the coders will code the full dataset using textual data analytic software that permits cross-classification and retrieval of transcripts and segments of text by theme. Key personnel will review the coding regularly to monitor quality. 3) Conduct descriptive and comparative analysis. Descriptive analysis will be conducted to identify the types, contexts, and nuances in the areas of investigation. The investigators will try to reveal patterns in the data, with issues mentioned repeatedly across interviews, and by certain sub-groups, suggesting normative perceptions or behaviour. 4) Develop conceptual framework of findings. Findings from the descriptive and comparative analysis will be developed into a conceptual framework that explains local constructs, the context and triggers of child and forced marriage and the conditions when they are accepted, normalized or justified. Both supporting evidence and counter-evidence will be used to test the conceptual model, which will be used to contextualize and interpret survey findings. 5) The data will be triangulated from all qualitative and quantitative data sources.

   Ethical considerations This study will be guided by the WHO recommendations for ethical considerations in researching violence against women and the CIOMS International Guidelines for Ethical Review of Epidemiological Studies (2002). Specifics of this have been presented below.

   The selected sample will be contacted physically using information collected during enumeration. The participant will be interviewed in private and will be informed orally of the purpose and nature of the study, its expected benefits, and voluntary nature of participation. As part of the consent procedure, the participant will be informed that the data collected will be held in strict confidence. To ensure that the participant is aware that the survey includes questions on highly personal and sensitive topics, the interviewer will forewarn the participant that some of the topics are difficult to talk about. The respondent will be free to terminate the interview at any point, and to skip any questions that she does not wish to respond to. Because of the low levels of literacy and concerns regarding confidentiality the interviewer will request verbal consent of the participant to conduct the interview. In order to interview minors (aged below 18) assent from guardians would be sought. Oral consent from a guardian will be sought first introducing the study as a study of health and life experiences of adolescent girls. Then the full consent procedure will be followed with the selected girls in question. The interviewer will then record on the consent form that the consent procedure has been administered, and note whether permission to conduct the interview has been granted.

   Participation in the study will be on a voluntary basis. No inducements will be made. The participant will be told that s/he is free not to participate in the study; to terminate the interview at any point, and to skip any questions that s/he does not wish to respond to.

   All interviewers will receive strict instructions about the importance of maintaining confidentiality. No interviewer will conduct an interview in their own community. No names will be recorded on the PDAs. Instead, all the study participants will be given a unique code and all the identifying information will be kept in a separate file. The file will exclusively be accessed by the researchers and will only be used for identifying the participants during endline. The de-identified data will be analyzed and care will be taken to present the research findings in sufficiently aggregated form to ensure that no study participant can be identified.

   During data collection and at the group level the intervention and the study will thus be introduced as activities focused on addressing adolescent girls' health and life experience to increase the acceptability of the study.

   However, in order to conduct the interview smoothly and for ensuring support for it the survey team/interviewers will explain the survey to any family member who is positioned higher in relationship hierarchy (if present) without revealing main focus of the study.

   Interviews will only be conducted in a private setting of participants' choice. If confidentiality cannot be maintained for any reason the interview will be rescheduled (or relocate) at a time (or place) that may be more safe or convenient for the respondent.

   Interviewers will be trained to terminate or quickly change the subject of discussion if an interview is interrupted by anyone. During the interview, the interviewer will forewarn the respondent that she will terminate or change the topic of conversation if the interview is interrupted, and will be able to skip to these questions at any point if needed. To ensure that interviewers gain experience about how to handle interrupted interviews, their training will include a number of role play exercises simulating different situations that they may encounter.

   Interviewers will be trained to be aware of any adverse effect and, if necessary, will terminate the interview if the effect seems too negative.

   Care will be taken when designing the questionnaire to try to carefully and sensitively introduce and enquire about different sections of the questionnaire, starting from the less sensitive one to more sensitive sections.

   Each interview will aim to end in a positive manner , which provides the participant with a positive outlook and reinforces her coping strategies. The questionnaire will include scripted conclusion for the interview stressing the importance of the information that she has provided, making comment on the respondent's strengths, and highlighting the unacceptability of her adverse experiences.

   Extensive training will be provided to the survey team. The training will not only discuss survey techniques, but also ethical issues, how to respond to, and if necessary, provide support to the participants.

   Mechanisms to attend to researchers' and field workers' needs A number of mechanisms will be adopted to attend to the needs of researchers and field workers. During the training process these issues will be openly presented, and participants will be given the option of withdrawing from the project without prejudice. During the research regular debriefing meetings will be scheduled to enable the research team to discuss what they are hearing, their feelings about the situation, and how it is affecting them. These meetings will aim to reduce the stress of the field work, and avert any negative consequences. This strategy proved effective in the Bangladesh component of the WHO multi-country study conducted by icddr,b.

   Despite these measures, some field workers may need to be given less emotionally taxing tasks, be given a break from the study or to withdraw from the research altogether. To account for these possibilities, sufficient numbers of field workers will be recruited to allow for a 10% attrition rate of interviewers over the study.

   The survey findings will be disseminated in a scientifically rigorous manner. Particular attention will be paid to ensuring that the findings are not used as a means to describe one setting or group as being "worse" than another.

Inclusion Criteria:

For girls' and boys'

• Aged 12-15 years
• Unmarried
• Usual resident of the community

For Community Members

• Aged 25 years or more and usual resident of the community

For fathers' and mothers'

• Father (or mother) of adolescent girl or boy group member

Exclusion Criteria:

For girls' and boys'

• Aged <12 or >15 years
• Married
• Not a usual resident of the community

For Community Members

• Aged <25 years or not a usual resident of the community

De-identified data will be shared according to icddr,b's official policy and as per requirement of the approved protocol.

After three years of the completion of the study

The data could be shared on request from researchers who will use data absolutely on non-profitable purpose. The request has to be reached to Ms. Armana Ahmed (aahmed@icddrb.org), Head, Research Administration, icddr,b. The icddr,b research administration will then review and decide about sharing de-identified data.