Misperceptions of Inequality and Sectarian Politics: Experimental Evidence from Lebanon

We conduct a randomized survey experiment with a representative sample of 3,300 adults in Greater Beirut, Lebanon. Respondents are randomly assigned to view an informational video that combines statistical evidence with personal narratives to convey two key messages: while overall income inequality in Lebanon is high, inequality between religious groups is relatively modest—meaning that Christians, Sunnis, and Shias experience similar levels of economic hardship. To evaluate the impact of this intervention, we use a combination of survey measures, a conjoint experiment, and behavioral outcomes co-designed with local nonsectarian NGOs to capture real-world expressions of anti-sectarian political engagement. We assess whether correcting misperceptions about intergroup inequality reduces sectarian political preferences, strengthens cross-sectarian class identification, and increases support for redistributive economic policies.

2025-04-24

2025-07-31

We begin by measuring first-stage effects on respondents’ beliefs about income inequality—both overall and between sectarian groups. We then assess downstream treatment effects on attitudes and behaviors across two domains: (1) political and civic engagement, with a focus on support for non-sectarian alternatives, and (2) pro-poor redistributive preferences and behaviors.

Beyond standard attitudinal survey measures, we partnered with local civil society organizations to design three behavioral outcomes that capture real-world political engagement: (i) registering interest in a non-sectarian NGO that delivers social services to low-income Lebanese across sectarian lines; (ii) applying to a three-month civic education program aimed at empowering citizens and fostering a new generation of anti-sectarian advocates; and (iii) enrolling in a workshop on government accountability and anti-corruption.

In addition, we assess treatment effects on candidate preferences using a conjoint experiment, enabling us to test whether correcting misperceptions about intergroup economic inequality increases support for anti-sectarian political parties and candidates.

In addition to the primary analyses, we use survey data to explore three potential mechanisms through which belief correction may influence political and civic behaviors: (1) Emotional channels — increased empathy toward the poor and heightened anger toward elites; (2) Social cohesion — greater trust, tolerance, and prosociality toward religious outgroups, as well as stronger support for the social contract; (3) Cross-cutting identities — stronger identification with economic class and national identity, relative to sectarian identity.

In the treatment arm, respondents are randomly assigned to watch an informational video midway through the survey. The control group proceeds without any video exposure. The video conveys a simple but empirically grounded message: while overall income inequality in Lebanon is high, income differences between the country’s major religious groups—Sunnis, Shias, and Christians—are relatively modest. That is, inequality in Lebanon is primarily within religious groups, not between them.

The video is composed of two parts, combining quantitative and qualitative evidence to target both analytical and emotional mechanisms of belief formation.

Quantitative Evidence: The first segment presents data from the Gallup World Poll and Human Rights Watch showing that, although income inequality in Lebanon is substantial overall, average income levels among Christians, Sunnis, and Shias are strikingly similar. Members of all three sectarian groups are distributed across the entire income spectrum. Visual aids highlight that economic hardship is widespread and not concentrated in any single group, and that intra-group inequality far exceeds inter-group inequality.

Qualitative Evidence: The second segment reinforces this core message through emotionally resonant, narrative-driven content. Drawing on recent evidence that storytelling can produce stronger belief change than statistics alone, this portion depicts three fictionalized young men—each from a different religious community—grappling with similar economic struggles, including underemployment, long working hours, and wages insufficient to support their families. This dramatized format is designed to cultivate empathy, reduce affective distance between sectarian groups, and emphasize shared hardship in a humanizing and relatable way.

The video was produced in collaboration with a professional film production organization that also operates as a Lebanese non-profit dedicated to promoting responsible filmmaking and supporting local directors and screenwriters. For the qualitative segment, the team cast local actors to portray the fictional characters, ensuring that the narratives were culturally grounded, contextually authentic, and emotionally resonant.

The randomization is done by a computer.

The randomization unit is the respondent.

No

N/A. We are studying individual-level randomization.

3,300 respondents.

Half of the sample is randomly assigned to the treatment group and views an informational video midway through the survey. The remaining respondents are assigned to the control group and complete the survey without any video exposure.

See PAP