Perspectives and preferences of both refugees and hosts on refugee local integration in Uganda

The study aims to assess the perspectives and preferences of both refugees and host communities regarding refugee integration in Uganda. Specifically, the objectives are:
1. To determine the average preference of refugees and host community members (Ugandans) for a more integrated refugee-host community.
2. To analyze the relationship between reported intergroup contact and the integration preferences of both refugees and hosts across various integration dimensions.
3. To compare the average preferences for integration scenarios between refugee and host (Ugandan) respondents.

The study will take place in Kampala and Isingiro districts (particularly around Nakivale refugee settlement). Data will be collected from both refugees and host community members (Ugandan nationals) living near the refugees, in both rural settlement areas and urban towns. Participants will be randomly selected using preexisting sampling frames from the Uganda Bureau of Statistics (UBOS), employing a two-stage cluster sampling strategy with the Probability Proportionate to Size (PPS) approach (with replacement). The final sampling stage may utilize satellite imagery or a listing exercise, depending on field and budgetary constraints. Based on power calculations, the study aims to include 700 participants, evenly split between Ugandans and refugees or asylum seekers. Data collection is planned to start in July 2024 and conclude by the end of August 2024.

The study features a fully randomized choice- and rating-based conjoint experiment embedded within a household survey. Participants will complete a survey questionnaire and be randomly exposed to a conjoint experiment involving six integration dimensions and their corresponding pairs of integration scenarios/community characteristics. The dimensions are: location freedom, political expression freedom, labor market freedom, integrated services, cultural fluidity, and citizenship. Each dimension's community characteristics will be illustrated with carefully designed pictorial representations. In five rounds, respondents will be presented with randomly compiled pairs of communities and asked to choose their preferred community based on the selected characteristics from each integration dimension. The survey will be programmed to randomize the community characteristics within each dimension and the order of the dimensions between respondents to mitigate ordering effects, while keeping the order of dimensions constant across the five choice rounds for each participant to minimize survey fatigue.

2024-07-10

2024-08-24

The primary measures of interest are;
1. A measure of the level of preference of both refugees and Ugandans for refugee local integration in Uganda
2. A measure of the importance of intergroup contact in shaping the preference for refugee local integration in Uganda

The level of preference for refugee local integration will be measured using a conjoint experiment. In the choice-based conjoint, participants will choose between two community options in response to the prompt: "Imagine what refugee-host communities would look like. You will see two possible communities, Community 1 and Community 2. Look at the images and listen to their descriptions carefully. Which refugee-host community would you prefer living in?" Each community option will include six integration dimensions with randomly selected sets of community characteristics, representing varying degrees of integration based on the combined characteristics. In the rating-based conjoint, participants will indicate the importance of their choice by answering, "How would you rate each of these community options on a scale from 1 to 7, where 1 means 'you absolutely dislike' and 7 means 'you absolutely like'?"

To measure intergroup contact, a constructed variable will be used to define substantial forms of interaction between groups, considering both intensive and extensive social contact. This variable will be created based on responses to four questions: (1) whether the respondent has shared a meal with an outgroup member, (2) how often the respondent has had casual conversations with an outgroup member, (3) how often the respondent has had business exchanges with an outgroup member, and (4) whether the respondent has worked with an outgroup member. Questions about interactions in the past month will capture the intensive margin of intergroup contact, while those about interactions over the past twelve months will capture the extensive margin. This constructed index will classify respondents as having either high or low intergroup contact.

The secondary outcomes are;
1. The mechanisms through which their preferences are formulated.
2. The closeness of their preferences to their reality.
3. The influence of some demographic characteristics of preferences for refugee local integration.

The mechanisms behind respondents' preferences will be assessed by having them rate the following statements for each community option on a scale from 1 (absolutely disagree) to 7 (absolutely agree): i) This community would be peaceful/safe, ii) It would be easy to work, start a business, or find a job in this community, iii) This community would have improved services, iv) Trust among members in this community would be high, and v) This community’s rules/laws would be inclusive/favorable for all.

The alignment of their preferences with their reality will be assessed by asking participants to compare the community options to their actual situation, responding to the question: "Which of these community options is closest to your current situation?"

The influence of certain demographic characteristics on preferences for refugee local integration will be examined by gathering data on the following factors: location (urban vs. rural), gender (male vs. female), employment status (employed vs. unemployed), education level (secondary vs. below secondary), humanitarian index (high vs. low), and both self-assessed individual and household economic status (non-poor vs. poor).

The study involves two types of conjoint experiments and subsequent analyses: choice-based and rating-based conjoint experiments.

In the choice-based conjoint, respondents will be presented with pairs of communities containing randomly selected sets of integration scenarios across six dimensions. Each community option represents a more or less integrated refugee-host community based on the combination of characteristics within each dimension.

Respondents will make choices for five rounds, with characteristics randomized within each dimension to minimize survey fatigue. The order of integration dimensions remains constant across rounds for each respondent, but it is randomized between respondents to avoid bias. This randomization and ordering are automated within the survey system.

In the rating-based conjoint, respondents will rate each community option on a scale of 1 to 7, indicating their preference. This rating process occurs five times, following each choice round.

Each community characteristic will be described alongside a corresponding pictorial representation. Images for two community options will be presented side by side to aid respondent choice and rating in each round.

Not available

Computer-based

The unit of randomization is the Individual.
I stratify by refugee status (refugee or Ugandan), and location (Kampala or Isingiro).

No

The treatment is not clustered. The planned number of respondents is 700 individuals; 350 refugees and 350 Ugandans, approximately equally distributed between Kampala and Isingiro district.

Given the structure of the conjoint, the planned number of observations is 7000 observations in the dataset from a planned total sample of 700 respondents. In a conjoint experiment, the number of observations in the dataset is equivalent to the effective sample size. Since each respondent will view two (2) community profiles each round, for five (5) rounds in total, the effective sample size is (2*5)*700=7000. From each respondent, we shall have 10 observations, one for each community profile per choice round. Note that the planned total number of observations is above the minimum as determined by the power calculations. This is to account for possible failure to reach and/or non-response given the refugee-host context in Uganda.

Conjoint experiment. No treatment arms.

The power analysis, based on Schuessler and Freitag (2020) and using their R package (Freitag & Schuessler, 2020), indicates that to detect Average Marginal Component Effects (AMCEs) of 0.05 in a conjoint experiment with two community profiles at a 5 percent significance level, an effective sample size of 3150 observations is required to achieve 80% power. This ensures that the probability of Type S errors (incorrect sign) is 0% and the exaggeration ratio (Type M error) is approximately 1.13. Note that the median of published AMCEs is reported as 0.05 by Schuessler and Freitag (2020). The effective sample size refers to the total number of observations in the dataset. To calculate the actual number of respondents, this effective sample size is divided by the number of profiles each respondent will evaluate. For a conjoint with an effective sample size of 3150, where each respondent chooses between 2 profiles across 5 rounds, the number of respondents needed is 3150 / (2*5) = 315 respondents. Therefore, with an effective sample size of 3150, I need 315 respondents, as each will view five pairs of community options. To achieve a balanced sample of refugees and host respondents, I would thus have to recruit approximately 630 participants in total. However, I aim for a total sample of 700 respondents to account for potential non-response, incomplete surveys, and attrition. Reference: Schuessler, J., & Freitag, M. (2020). Power analysis for conjoint experiments. https://osf.io/preprints/socarxiv/9yuhp/