Experimental Design
The sample will be randomly divided into either a treatment or a control group. We will measure trust at baseline for all participants using a questionnaire.
The questionnaire will be divided into two sections. In the first section, we will ask questions about trust in different institutions, especially in trust in the police. In the second section, we will ask demographic questions (age, education).
To keep track of participants, we will ask for the respondent's last 3 or 4 digits of their national ID number. This could serve to identify the respondent at baseline and endline, as well as at the FOSEP registry, and we will keep record of the address in which the interview was performed. We will go to the same address for the endline survey. The identifying information will be collected using pen and paper, and will never be matched with the participants' answers.
Regarding the treatment, this will consist in the provision of information about the FOSEP of the Comuna of the respondent. First, we will add to the questionnaire a question about whether the respondent knows about the FOSEP program and, if yes, whether he/she thinks it is a positive program. Second, we will handle the respondent a small information sheet about the next FOSEP in their Comuna, with the date and location and some information about the context of the experiment.
We will use the treatment to perform an intent-to-treat analysis. Our hypothesis is that the treatment will increase participation in the FOSEP and in turn, will lead to a change in trust in the police.
Once each Foro de Seguridad is over, we will return to each surveyed household or business to conduct the endline survey. In this part, we will again measure each respondent´s trust in the police. The aim is basically to measure the difference before and after the intervention's implementation in the difference in trust in the police between the treatment and the control group.
Interestingly, we will be able to perform an analysis of the diffusion of information, because the respondent in the household or business we will interview might change between the baseline and endline surveys.
Due to the fact that the treatment and the control group will be selected at random from the sample, any difference in the difference in trust in the police between TS and CS can be attributed to the intervention. The same is true for TO and CO. We will therefore have two main outcomes: the difference in the difference in trust in the police between TS and CS, as well as the same measure for TO and CO. Any result different from zero for this second outcome can be attributed to diffusion effects within each household or business.