Social interactions and the integration of internally displaced people in Mozambique

This intervention will consist promoting in gathering internally displaced people (IDPs) and local hosts in four community meetings, spaced by two months. In each community meeting, a specific pre-defined topic will be addressed. Community meetings will be moderated by a local religious leader. Individuals in the control group will not participate in any community meeting.

The empirical assessment will include a baseline assessment before the first meeting. Every two months, immediately before each subsequent meeting, a follow-up assessment will be conducted. Two months after the last community meeting an endline assessment will be conducted. Each assessment will include surveys, list-experiments and an implicit association test. In addition, the endline assessment will also include lab-in-the-field games.

Community meetings will follow the framework of a public dialogue, developed by Herzig and Chasin (2006). The first community meeting will cover topics related to the arrival and integration of IDPs in local neighborhoods, as well as other issues related to local daily life. The main objective is that participants get to know each other and to recognize the challenges each other group faces (IDPs / locals). This meeting will also be an opportunity for IDPs to tell their stories, and for locals to learn from IDPs.

The second community meeting will focus on prejudice, and it will follow a prejudice exercise that was also developed by Herzig and Chasin (2006). The proposal of this exercise if for meeting participants to reflect on the prejudices felt in their daily lives in the neighborhood, and that come from the individuals in the other group (IDPs / locals).

The third community meeting will focus on religious tolerance, leveraging on the theological knowledge of local moderators. The objective will be to clarify some doubts individuals may have about Islam – given the intervention is being conducted in the context of an ongoing religious extremist insurgency.

The fourth meeting will lead participants to reflect on what they learn throughout the community meetings program, specifically on what can be useful throughout their lives. Participants will also discuss aspirations, and will be encouraged to keep holding community meetings (after this intervention is concluded) if they wish to do so.

Each assessment (baseline, follow-ups and endline) includes several components. All assessments will include a survey measuring demographic characteristics, trust, attitudes and beliefs and social networks. Open ended questions will also be included, with the objective of using text analysis to study individuals thoughts, sentiments and emotions towards IDPs, locals and insurgents.

In addition, all assessments will also include 3 list experiments. One will cover IDPs sense of being discriminated by locals (if the respondent is an IDP) or discrimination against IDPs (if the respondent is a local). The second list experiment will capture individuals trust in people belonging to the other group (IDPs / locals). The third list experiment will capture individuals’ preference for insurgents, using a graphical format where pictures will be displayed – instead of reading out loud multiple sentences (as in the other “traditional” list experiments).

Every assessment will also measure implicit bias towards insurgents by conducting an implicit association test. The two bias categories will be “insurgents” or “government authorities”. The implicit association test will display images (insurgents, government, happy faces, sad faces). As most participants will not be familiar with implicit association tests, a training test (using snakes and food, instead of insurgents and government) will also be conducted.

With the exception of the baseline assessment, all follow-up and endline assessments will also use survey questions to track individuals’ social networks throughout the intervention. The objective is to measure whether community meetings are positively affecting participants’ social networks.

The endline assessment will also include three lab-in-the-field games: dictator, public goods and joy-of-destruction. Participants will play in pairs, and they will not be told their opponents’ identity. Instead, participants will only be informed whether the opponent is an IDPs or a local.

Throughout the intervention, spillovers will also be measured using two strategies. The first one will consist on measuring effects based on geographic proximity to treated individuals. The underlying assumption is that proximity to individuals who participated in community meetings is more likely to generate spillover effects.

The second strategy will consist in inviting participants’ close contacts to a short survey at baseline and endline. Specifically, the main study participants will allowed to invite 2 persons of their choosing to answer a short-survey at baseline. In this survey, invitees will answer questions measuring beliefs and attitudes. At endline invitees will return and answer the same survey. The underlying assumption is the treatment effects may spillover among individuals who interact more with participants of community meetings.

2023-04-25

2024-01-31

a) Religious tolerance (measured using an index composed of survey questions; implicit association test score; graphical list experiment; text analysis + open ended questions)
b) Tolerance towards IDPs (measured using an index composed of survey questions; list experiment; text analysis + open ended questions)
c) Well being of IDPs
d) Trust in IDPs / locals

The endline assessment will also include lab-in-the-field games measuring preference for the other group (dictator game), cooperation (public goods game) and anti-social behavior (joy of destruction game)

Religious tolerance index: constructed by averaging survey questions measuring preference for freedom of religion, imposition of the Sharia.

Religious tolerance - Text analysis + open ended questions: the answers to open ended questions asking about individuals’ thoughts about respondents will be the target of an emotions and sentiment analysis.

Index of Tolerance Towards IDPs: constructed by averaging survey questions measuring locals’ preference for IDPs staying in local neighborhoods or Pemba. Beliefs and attitudes towards IDPs.

Institutional Trust Index: Trust in different individuals will be measured directly in the survey (i.e. the level of trust that a participant has in family members, neighbours, IDPs, locals, political leaders, etc.). I will also construct a trust index by aggregating participants' trust level in different individuals: a) general trust index (averaging all questions); b) government, c) local people, etc.

a) Aspirations (measured using survey questions)

b) Social networks (measured using survey questions)

Aspirations index: composed of survey questions capturing individuals assessment on future income, occupation, and children level of schooling.

Social networks: connection index aggregating individuals personal networks, using survey data.

The study sample will be composed of both internally displaced people (IDPs) and locals. Individuals in each of these groups will be randomly selected to participate in the study.

This study will take place in 9 (out of 13) neighborhoods in Pemba. In each neighborhood, a random sample of IDPs and locals will be selected. IDPs and locals randomly selected to be in the study sample will be randomly allocated to cohorts of 14 people (8 IDPs and 6 locals). Cohorts can either belong to “treatment” or “control” groups. Individuals in the treatment group will engage in four community meetings, spaced from each other for about two months. Individuals in the control group will not participate in any community meeting.

With the purpose of measuring spillover effects, study participants will be allowed to invite two persons close to them to participate in a short survey at baseline and endline. In addition, coordinates of the homes of all study participants will also be collected.

The selection of IDPs and locals into the study will follow different procedures.

IDPs will be randomly selected based on the Displacement Tracking Matrix, built by the International Organization for Migrations (IOM). This dataset includes the names of all IDPs heads of households, telephone numbers and detailed location within the neighborhoods. The access of the research team to this dataset follows a data sharing agreement with IOM.

The DTM is stored as an Excel file. Therefore, the research team randomly sorts the names to be invited to participate in the study. IDPs that cannot be found are replaced by the names following the list constructed by the research team.

Regarding locals, there is no roster such as DTM. Instead, we will use the method of random walks. The sampling interval will be based on the most recent official information, which comes from the 2017 Census of Mozambique.

There will exist 2 different randomisation strategies: a) IDPs; b) Locals.
a) IDPs being invited will be randomly selected based on the roster of displaced people, kept by IOM. In each neighbourhood 68 individuals will be randomly chosen out of all IDPs in the roster. Specifically, this will be done by using a computer to randomly pick 68 individuals among everyone in the roster.

b) Locals's invitation in each neighbourhood will follow the method of a random walk. Enumerators will have the help of local guides in order to find the center of the neighbourhoods. From there, they will move in different directions, randomly picking up local households according to the designated sampling interval (neighbourhood-specific).

Individuals selected to be in the study sample will then be randomly assigned to treatment or control cohorts. Treatment assignment will be done by lottery, at the moment when individuals are being invited to be study participants. Invitees draw a card from a bag. Each card has a different number, which indicates to which cohort the invitee will be allocated throughout the entire study.

Individual level

Yes

9 neighbourhoods. Each neighbourhood is planned to have 8 cohorts (clusters), each one including 8 IDPs and 6 locals (totalling 14 invidivuals per cluster)

The target study sample if 1008 individuals: 576 IDPs and 432 locals. The different proportion of IDPs and locals is explained by a higher predicted attrition rate among IDPs throughout the intervention (based on evidence from a previous assessment).

Each neighbourhood will have control and treatment clusters. I estimate 4 treatment clusters and 4 control clusters (each cluster is planned to be composed of 14 individuals, ideally 8 IDPs and 6 locals).