Home IGIER  


Evidence from a Randomized Campaign

 Chad Kendall          Tommaso Nannicini*          Francesco Trebbi

The experiment in a nutshell

We study the causal impact of campaign information on electoral outcomes and voters’ beliefs about political candidates. To do this, we perform a large randomized controlled trial conducted during a mayoral campaign in 2011 in a medium-sized Italian city, Arezzo. Voters received hard and verifiable information via mail (campaign flyers) or via phone (audio messages) about the valence or/and ideology of the incumbent (Giuseppe Fanfani). The informational treatment was administered as part of the incumbent’s actual campaign and covers the entire voting population.

The city was divided randomly into four areas:

-    Voters in the first area received information on valence
-    Voters in the second area received information on ideology
-    Voters in the third area received information on both
-    Voters in the fourth area received no message.

In the week before election day:

-    100% of families received mailers designed by professionals
-    25% of families received phone calls by volunteers (no robo call, volunteers talk with voters for about two minutes), ending with recorded message by the candidate.

Furthermore, the incumbent provided the information for each message and we let him choose between two alternative ideology messages.

Surveys were administered before and after the treatment. They capture individual voters’ beliefs about valence and ideology with respect to both the incumbent and the challenger.

Download the paper

Download the Presentation

Text of the First Survey


Campaign Flyers

Campaign Phone Calls
Text of the Second Survey


Both surveys were realized by IPR Feedback


* Nannicini acknowledges financial support from the European Research Council (under grant No. 230088)

Last updated July 29, 2013