Perceived Inequality Across Ethno-Religious Groups: Experimental Evidence from Lebanon

We will evenly split the sample into a treatment and control group, stratifying by religious group. We will show to the treatment group a video on income inequality across religious groups, conveying that income inequality between Sunni, Shia, and Christians is low, and that income inequalities are much more pronounced within each religious group than between them using data from the Gallup World Poll, World Inequality Lab, Assouad (2023) and UN-ESCWA. To reinforce the message underlying our treatment arm, the video will also include qualitative testimonies by Lebanese citizens from different religious backgrounds sharing their common economic difficulties. The survey questionnaires will be the same across treatment status. The video appears mid-way through the survey, after the baseline variables have been collected.

2024-06-01

2024-10-31

We will first document changes in beliefs about inequality and anti-sectarian views (first-stage) and to what extent incorrect beliefs vary based on four factors emphasized in the literature: (i) exposure to sectarian rhetoric proxied by self-reported exposure to sectarian media outlets3, (ii) income level and dependence on clientelism to meet basic needs, (iii) network diversity proxied by the quantity and quality of intergroup contact (Lebow et al. 2023; Breza et al. 2020). We will then test whether correcting misperceptions can: (1) increase social
cohesion and solidarity between low-income households belonging to different religious sects using self-reported attitudes towards outgroups including intergroup trust, comfort in living in mixed neighborhoods, willingness to interact and work with outgroup members; comfort with having a relative marrying outgroup members; (2) increase the salience of class identity and national identity and decrease the salience of religious identity; (3) increase support for redistribution and pro-poor policies (e.g., minimum wage, national poverty targeting program, income tax rates, and flat capital income rate) and tax morale; (4) decreases support for sectarian institutions and policies. Our outcomes will mostly be measured using survey data following the relevant literature (Okunogbe 2018; Blouin and Mukand 2019; Depetris-Chauvin et al. 2020; Mousa 2020; Alesina et al. 2021; Hoy 2023). However, we plan to measure (4) using survey-based measures of attitudes (e.g., attitudes toward sectarian politics and acceptability of voting for politicians from other religious groups) and direct behavioral measures, including participation to meetings to learn about alternative political candidates and concrete actions citizens can take to oppose sectarianism.

We will implement a survey with a representative sample of adult residents of Greater Beirut. We will recruit a minimum of n=3,300 participants from the three largest religious groups in the country: Christian, Sunni, and Shia, each composing roughly one third of the sample. Furthermore, the sample will be representative in terms of neighborhoods and gender. We will assign half of the respondents to watch a video on income inequality across religious group, and half to a pure control group. The randomization will be stratified by neighborhood, gender, and religion.

Not available

Done by survey company using computer

Individual

No

None

N=3300

Treatment: 1650
Control: 1650

We will estimate effects using an intent-to-treat framework, regressing outcomes on treatment dummies. We have conducted power simulations for attitudes toward other religious groups, assuming effect sizes in line with Alesina et al. ‘s (2021) study of perceptions of racial gap. We need a sample size of 3,300 respondents to detect an effect size of a similar magnitude. We intend to conduct a logistics pilot with 50 respondents to ensure sufficient variation on all the survey questions and to test different formats for the messages in the video interventions (e.g., combining qualitative and quantitative information). Based on the pilot results, the research team will adjust the survey and select the intervention's most effective and easy-to-understand video format.