 Thank you very much. Good morning. It's a pleasure to be here. Thank you very much for having me. I'm going to talk about a project that we conducted. This is joint work with Alex Armand and Ines Villela, who are sitting in the audience. This is a project on conflict prevention in northern Mozambique. So let me start with a little bit of motivation. In recent years, most of the major violent conflicts have happened in Muslim majority countries. And Islamic radicalization is one of the main correlates of violent conflict in the world today. So the question that I'm going to target today is, can Islamic radicalization and related violence be countered through campaigning by local religious organizations? So we follow a campaign against violence sponsored by religious organizations in Cape Delgad, Mozambique. As mentioned in the previous session, Cape Delgad has seen a violent insurgency since 2017, led by Islamic radicals. There is also link to ISIS, which resulted in more than 3,000 deaths and 800,000 refugees to date. The campaign was sponsored by different Islamic organizations and focused on the key message that very simple message that religion is not violence. The campaign was broadcasted through community radios and individual voice messages. These were sent to cell phones. And so we follow a radio treatment and the message treatment, which were randomized. So this is an experiment. We measure impacts on survey attitudes and various behavioral measures, including on a disocial behavior and trust. Let me also tell you that this is the first time this project is being the results are being presented or some of the results. So this is preliminary. So in terms of a preview of results, what we find, in case I don't have time and I probably will not have time for everything here, the radio treatment we find improves attitude significantly, making people less supportive of violence and of extreme social views related to radical Islam. We don't find effects on antisocial behavior from this radio treatment, but we find that the difference between in this antisocial behavior that we are able to measure, we find that a difference between harming Muslims and others increases. So these difference increases. The belief that more antisocial behavior will arise also increases because of this radio campaign. The radio treatment decreased coordination between individuals from different religions. We don't find significant effects of the message treatment on cell phones, except for an effect on perceiving Muslims positively. We don't find complementarity between the radio and the message treatments. So I'm going to skip the literature because I don't think I will have time. Let me just remind everyone about the context here. There was a substantial discovery of natural gas in Khabdelgad starting in 2010. Khabdelgad is remote and primarily rural, with high poverty and child mortality rates, even for national standards. Conflict started in Khabdelgad at the end of 2017, as I mentioned before. There are many associations with radical Muslims, but I must say that most perpetrators are Mozambican. So the treatments we follow can be described as follows. These were developed by the two main Muslim authorities in Mozambique, Cizlamo, the Islamic Council of Mozambique and the Islamic Congress. We also partnered with the Christian Council of Mozambique. So these three together, they compose the coalition for this campaign. So the treatment consisted of interventions by the religious leaders of each group, supporting peace and explaining how their religion supports peace. And so specifically, you have a few quotes that are basically the messages that were conveyed to in the radio and also in the voice messages on cell phones. So the final message, religion is love and friendship. Together we shall walk to achieve peace. So in terms of the radio campaign, the radio spots were aired in eight community radios spread around Khabdelgad. There were daily broadcasts between June and July 2021 and two airings per day between September and November 2021. Both Portuguese and local languages were employed in the airings. In terms of the voice message treatment, this is a partnership with a private provider experienced in delivering information campaigns. So each individual in our treatment sample received four voice calls during this period with approximately two weeks in between. And basically the voice calls were had the same content I've just described before. So from the various religious leaders. So the messages were sent from September to November 2021. The languages that were employed were those of the listeners. The campaign had a high success rate, 87% picking up the phone on average of the four rounds. In terms of sampling, I'm not going to go over the full detail here. Let me just mention that our sample is based on a previous study that we conducted, which had a representative sample of Khabdelgad, of the villages of Khabdelgad with 206 villages. This study was a study that was conducted before 2017. So just before 2017. We keep all these villages but satisfying two criteria. First, having reliable phone coverage. And second, still existing after the attacks started in 2017. From these, we reached a substantially lower sized sample with 146 villages. So this is the sample of villages that we take in this study. Within these villages we were also using the initial sample of individuals that we had from the previous study. But we enlarged this sample through phone contacting the original sample or the people that we were able to reach in the original sample. And we enlarged this sample to a sample of 1400 individuals. So following the, let me mention a little bit about radio coverage here because that's going to be instrumental for the research. And the voice message randomization that we follow. In terms of radio coverage, we follow the literature. And we construct topography corrective radio coverage using a model, which is the long lay rise irregular terrain model. So this model takes in station and antenna parameters, which we gathered from the stations, the radio stations, and topographic characteristics in Carbel del Gard to determine which areas receive a signal from the station and at what strength. So the assumption that we are going to follow here is that there is quasi random variation for the radio treatment, which is to say conditional on distance from the antennas. The presence of obstacles between the antenna and the listener is exogenous to the listener's behavior. We are going to assume in the results I'm going to show a standard threshold for quality of radio signal, which is a specific international standard. For the voice messages, we randomized treatment at the level of the village. So this is pure randomization. And we used block randomization. So this is to give you a sense. This is Carbel del Gard province. And the blue areas are the covered areas in terms of radio signal from these eight community radios. The green spots are the 146 villages in our study. Now in terms of measurements, we employ two types of measurements. First survey data. So we conducted phone surveys with our sample. We had the baseline in May to June 2021, December 2021 up to February. This year we had the end line. We included in the survey standard demographic questions and also outcome questions relating to outcome variables, asking about support for extremist religious positions, namely related to violence, government and democracy, social norms relating to gender, and the use of photos. So in terms of behavioral data, we also collect these behavioral data in the context of these and specifically the online survey. So by phone. And we conducted a few standard activities. Namely, we have this anti social behavior measure that comes from a game that is called the joy of destruction in which two individuals can destroy each other's endowment at the price. Each participant is paired with two different individuals. One with a distinctive Muslim name and another with a distinctive non Muslim name. We also elicit beliefs about what the other person is going to do in this activity. And then we also play a standard trust game with two counterparts as well. Then we have a number of other behavioral activities in which we are able. I'm not going to go over. I don't have time for this, but just to give you a sense. So we have an activity on which is called the SMS activity in which people can pay to send a peaceful message in which we observe whether people really want to pay to send a peaceful message. So this is incentivized with value. We have a coordination activity in which participants first select their counterpart between two possibilities, a Muslim and a non Muslim. And then they need to coordinate to receive a price. We have a salience activity in which people hear a story with different characters, some with Muslim names, some others with non Muslim names after that they are asked about these characters. So this is a salience activity in the sense of memory. And finally, we have a perceptions activity which measures implicit prejudice against Muslims and non Muslims using an index of characteristics. I'm going to show results on this. I'm going to, this is very standard. We are going to estimate intent to treat effects basically using the two main coefficients here on the radio treatment and on the message treatment. So very quickly on the results, I'm going to skip the descriptive statistics very quickly. So we see effects of the radio campaign on decreasing support for religious violence, also increasing support for democracy and increasing support for gender equality. So we see basically these effects of the radio treatment. We don't see effects of the message treatment for these survey questions. For the destruction activity, we don't see effects of significant effects. Although the radio treatment has these negative coefficients, so decreasing destruction, we see an increase in destruction of Muslims versus non Muslims, which is something that we need to ponder about. We also see an increase in the belief about destruction from the radio treatment. We don't see effects of the message treatment here. Finally, with respect to the other behavioral activities, we see negative effects of the radio campaign on trust, although these are not significant. We see a decrease in coordination in selecting the other religion for the coordination activity. And we have some positive effects namely for the message treatment in terms of the perceptions activity I mentioned before. So just to conclude, the radio campaign was particularly effective at improving attitudes in the direction of less support for violence, more support for democracy, and less support for extreme social norms. Still, we have a few caveats here. The radio campaign led relatively more anti-social behavior against Muslims and a general belief that others behave or will behave in an anti-social manner. Coordination with individuals from different religions was also decreased, so we have to have in mind these caveats. We don't have clear effects of the voice messages despite the high degree of hearing of these messages. So overall, we have good indications, although again, this is preliminary evidence and not fully optimistic because of the caveats that I've mentioned regarding the mobilization of local religious organizations to counter Islamic radicalizations through the radio. Thank you very much.