Community conversations as a strategy to change harmful traditional practices against women

ABSTRACT This article conducts a series of list experiments to detect whether community conversations contribute to a change in thinking about harmful traditional practices in Ethiopia. While our findings are mixed, we provide evidence that community conversations are indeed a valuable instrument to induce a change in social values in order to empower women.


I. Introduction
According to DFID (2012), about one third of all women and girls in the world has been affected by one or another form of abuse. More than one quarter of the women report that they have been victim of sexual violence. Especially in developing countries, where violence towards women is one of the major reasons of missing women among older cohorts (Anderson and Ray 2010), female conditions are dramatic. Girls and women are treated differently than boys throughout their life, which manifests itself in less economic and political empowerment and limited control over their bodies. Many young women are trapped in equilibria where low economic and political empowerment and incomplete control over their bodies reinforce each other (Bandiera et al. 2014).
Because of the crucial importance of improving female positions in developing countries, reducing violence and abuse against women is high on the policy agenda. Indeed, a reduction in violence against women and young girls is currently one of the most important development goals worldwide.
Harmful traditional practices (HTPs) are among the major causes of discrimination against women in many societies. The term HTP is used for practices such as female genital mutilation, early female marriage and forced prostitution that affect the physical and mental health of the victims adversely. These practices are often deeply rooted in culture. Many nongovernmental organizations (NGOs) try to reduce HTPs by providing skills training, by setting up help posts and by organizing awareness campaigns. In order to improve awareness in traditional societies, several NGOs have also started to organize programmes of community conversations. Community conversations offer participants the opportunity to increase their knowledge on children's and women's rights as well as child protection, and to identify and refer cases of child abuse. The aim is to enhance awareness raising of community members on violence against girls and young women, to encourage discussion among community members and to address violence against girls. Thus, community conversations are meant to change social values, beliefs and attitudes with respect to HTPs against female, which hopefully will eventually lead to a reduction in violence against girls and young women.
This article reports the findings of an evaluation of a community conversation programme in Ethiopia. We focus on the key question as to whether the community conversation programme has contributed to a change in social values, beliefs and attitude on HTPs against women. Since HTPs deal with sensitive issues, our main evaluation methodology uses an indirect questioning method to detect truthful answering. To the best of our knowledge, we are the first who have conducted a series of so-called list experiments to detect whether community conversations contribute to a change in thinking about HTPs in the society. While our findings are mixed, we provide evidence that community conversations are indeed a valuable instrument to induce a change in social values in order to empower women.

II. The community conversation programme
We consider a community conversation programme in Ethiopia, conducted by an Ethiopian NGO. The NGO organizes community conversations for a group of approximately 60 members in the target Kebeles (a Kebele is an administrative unit in Ethiopia, akin to a neighbourhood) of the organization. Most of the participants are selected from different groups in the community because of their active role in the community and/or their believed ability to convince others. The facilitators try to include children in the community conversations to make the group representative of the kebele with an age range of 4-80 years old. The organization is responsible for organizing the programme, training the community conversation facilitators, providing coffee, tea and bread, dispersing correct and good information and providing certificates when the participants completed all sessions. The NGO has organized a total of 35 community conversation programmes between 2010 and 2013. Ten sessions are organized per group of participants starting in January until the tenth session is planned. From each Kebele 120 participants take part in the community conversations by the NGO each year. After participating in all ten sessions, approximately 80 receive their certificate.
The community conversations offer participants the opportunity to increase their knowledge on children's and women's rights as well as child protection. In addition, the participants are trained to identify and refer cases of child abuse. At the first meeting a facilitator provides initial topics, but the participants determine the specific topics that will be discussed based on the relevance of the issues in their Kebele. A selection of the topics that have been covered are children's and women's rights, health, creating jobs, child abuse, family planning, child labour, child trafficking and sexual abuse.

III. Do community conversations change social values?
The main purpose of the community conversations is to educate the community in order to prevent child abuse, create child protection and empower girls and women, mainly via a reduction in HTPs . The community conversations aim at creating awareness by distributing the information gained at the conversations to the larger community through the formulation of action points during the community conversations.
HTPs deal with sensitive issues, which may be difficult to detect with direct questioning as people may try to give 'socially' correct answers. Therefore we decided to use an indirect questioning method to illicit truthful answering. The method we used is known as a list randomization. A list randomization, 1 which allows the respondent to conceal the answer, works as follows. Half of a group of respondents is randomly assigned to a set of base statements, the other half is randomly assigned to the same set of base statements, plus a statement related to the sensitive issue. Next, individuals from both groups are asked to tell with how many statements they agree (not with which statement they agree). By doing so, the average amount of statements both groups agree with is obtained. Due to the random assignment of the 'sensitive question' to one of the groups, an estimate of the proportion of the sample that is engaged with the sensitive issue is given by the difference in the mean true statements of the two groups.
In order to examine the impact of the community conversations programme on values on HTPs, we conduct list experiments for two groups. The first group consists of approximately 200 individuals who had just completed an entire community conversation programme of the NGO; the second group contains about 200 individuals who were shortlisted for a new community conversation programme of the NGO. By comparing a group that has recently followed community conversations, with a group that starts a series of community conversations, we ensure that the groups are comparable and we avoid biases due to self-selection. Some simple balancing tests confirm that the two groups are comparable in terms of observables.
We randomly divide both groups into two groups, Group A and Group B. In total, we test 9 sensitive issues, which are given in Table 1. Both groups received the same 9 sets of 3 standard statements. Per set of questions either Group A or Group B is also assigned to one of the sensitive statements. Finally, groups A and B have been asked to report how many statements are true. A description of the set of base statements can be obtained on request. The full set of sensitive statements is presented in Table 1. Table 2 presents the proportions of people that agree with these sensitive issues, both for the group that has followed the community conversation programme (Treated) and the group that has been selected for a new round of community conversations (Control).
While the results are somewhat mixed, community conversations indeed seem to contribute to a change in awareness. It is encouraging that there are big differences between the treated group and the control group for questions 3 and 5, which indicates that community conversations help to change values and behaviour regarding how people think about the responsibility of being sexually abused (Q3: Women and girls who are being sexually abused are at least partially responsible themselves) and about decision making (Q5: A man should have the final word about decisions in the home). The community conversations also seem to affect bargaining power of females (Q6: A wife has a right to buy and sell things in the market without asking the permission of her husband) and thinking about prostitution (Q7: If there are no other options, prostitution is a reasonable choice). For the other sensitive statements we do not find evidence for a difference between the two groups. Yet, our list randomizations provide preliminary evidence that community conversations are valuable instruments in the process of reducing harmful traditional practices, i.e. practices and cultures which affect the health and well-being of women.

Disclosure statement
No potential conflict of interest was reported by the authors.

Funding
Part of this research has been financed by .107], the Netherlands. A wife has a right to buy and sell things in the market without asking the permission of her husband Question 7 If there are no other options, prostitution is a reasonable choice. Question 8 Girls should be circumcised Question 9 The practice of early marriage should continue