|Title:||Entertaining, informing, discussing: How do media spread messages of peace and violence?|
|Name:||Elizabeth Levy Paluck|
We test the causal effects of a democracy promotion radio intervention launched by an international non-governmental organization (INGO) in South Sudan. The serial radio episodes addressed topics that are relatively new to listeners, regarding democratic institutions and procedures such as anti-corruption law and elections. Other radio episodes addressed topics about which listeners likely have pre-existing positions, such as the role of women in politics and the meaning of citizenship. The INGO augmented the radio intervention with face-to-face, formally moderated listener group discussions regarding the program content. We randomly assigned these different aspects of the radio intervention ("treatments") in a hybrid lab-field study. Specifically, we randomly selected small groups of neighbors and friends in the capital city Juba, and randomly assigned these groups to experience one or more treatments (different radio episode topics, moderated discussion) over the course of a full day. We measure learning, attitude and behavioral shifts by comparing listening groups to no-treatment controls and to one another.
We find that encouraging discussion about the radio program has substantively and statistically significant effects on learning about the meaning of new democratic concepts and institutions. Discussion, and particularly discussion of the relevant radio episodes on democracy, improves attitudes toward democratic issues, but discussion of womens involvement in politics causes more negative attitudes toward these issues. We also find substantively and statistically significant behavioral shifts caused by discussion in terms of donating money to civil society organizations (CSOs) and actual reporting of corruption and volunteering with CSOs up to one month following the experiment. Explanations for the influence of this radio-inspired discussion are suggested by psychological theories regarding the social influence of listening partners. However, listeners' perceived partner agreement with the radio program has mixed effects; we only find an association between treatment effects and perceived partner agreement with respect to attitudinal persuasion. Finally, the gender of participants and the gender composition of listening groups conditions our average treatment effects in significant and relatively patterned ways. Women respond more positively to messages about women in politics and citizenship, and are less likely to be negatively affected by discussion about these topics. Listening to radio content about women in politics and citizenship also motivates greater behavioral responses among women and among listening groups that include women, in terms of donating money and requesting the contact information for CSOs. However, women are not more likely than the overall sample to donate to women's causes.