The experiment in a nutshell

Do political campaigns affect voters’ behavior? If yes, are positive or negative messages more effective in doing so? Does information about electoral candidates have an effect on voting behavior? These are the research questions that our experiment addresses.

Using an online panel randomized in three groups (1. positive campaign information;

2. negative campaign information; 3. no campaign information), we simulate the “entire” political campaign by Giuliano Pisapia, the main challenger of the incumbent mayor, Letizia Moratti, in the 2011 Milan municipal elections.

Online participants are exposed to campaign texts and videos from both candidates, but we randomize the texts and videos by Giuliano Pisapia based on their content.

•The "positive campaign information" group is exposed to texts and videos by Pisapia focusing on his own proposals.

•The "negative campaign information" group is exposed to texts and videos by Pisapia focusing on the incumbent's mistakes.

•The "no campaign information" (control) group is exposed to a placebo survey, that is, it does not receive any campaign information.

We then look at the effect of our informational treatments on several outcomes of interest (namely, a wide set of variables capturing voting behavior and general political attitudes). We also explore heterogeneous treatment effects across observable participants' characteristics, such as gender and education.