Memory of Corruption, Democratic Reforms, and Electoral Behavior in Paraguay
How to cite
Straub, Stephane. 2015. "Memory of Corruption, Democratic Reforms, and Electoral Behavior in Paraguay." AEA RCT Registry. AEARCTR-0000531. https://doi.org/10.1257/rct.531-1.0
Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.
* Primary investigator
Abstract
Paraguay has long been characterized by weak institutions, a very corrupt political class where large scale clientelism is the rule, and a low degree of public involvement, with widespread vote-buying at election time and low knowledge of- and trust in- electoral institutions. However, following the publication of information on public institutions payrolls, a number of high-profile corruption scandals involving members of Congress have recently erupted, spurring a strong public reaction contrasting with the usual apathy of citizens in the country, against both the corruption facts per se and against the widespread political culture of impunity.
These events provide the opportunity to follow and study in real time the emergence and development of political awareness in a quite new democracy, and to experiment how different democratic institutions may allow this political awareness to translate into effective political control.
More specifically, the project will combine surveys and experiments, to study how voters in Paraguay will respond in the future to this unusual high-profile event of political corruption (our initial “shock”), focusing on the psychological mechanisms through which they process and subsequently recover information about wrongdoings by candidates, and analyzing experimentally how different electoral institutions shape voters’ ability to punish/reward politicians and/or political parties.
Intervention
In the context of a survey with the household head and her/his spouse, covering detailed questions on demographics, interactions with different institutions (schools, health sectors, social networks), migration, sources of news, politics, corruption perceptions, and moral psychology aspects, we ask him/her to participate in a mock vote, replicating the April 2013 senate election.
Each participant is given a set of news extracts, including several pieces of information on politics, culture, and football. The control group receives a neutral treatment showing the current composition of the senate. The treatment group will receive a reminder of the corruption scandal in the senate, in two forms. Treatment 1 is a simple reminder of the facts. treatment 2 adds a reminder of how the facts spurred a reaction in social networks.
The culture and football news are common to control and treatment.
Subsequently, participants are asked to vote as if they were voting again in the 2013 senate election, based on a slightly simplified bulletin including the 4 main lists. They vote twice under two randomly ordered modalities, namely blocked lists (voters chose only one list without being able to alter the order of candidates), or open lists (voters chose only one list, but have the possibility to erase some of the names).
Experimental design
The experimental design implies first randomizing the treatment across households.
Before voting, household heads (member 1) will be exposed to the news package including either the control materials, or one of the treatment materials (the reminder of the corruption case). They will then be asked to vote under the two alternative modalities (closed or open lists) in random order.
The next step will imply within household randomization.
The spouse (member 2), will be asked to do the same, after being exposed to a news package determined on the basis of the one given to the household head. To ensure balance, if member 1 was assigned to the control group, member two will randomly be assigned to one of the treatments. If member 1 was assigned to one of the treatments, member 2 will be assigned to the control.
Given a balanced household level control / treatment 1 / treatment 2 randomization (proba 1/3 each group),the within household randomization implies an overall pattern of 50% of HH with C-T1, and 50% with C-T2.
Design details
Cited by
No citation data available for this registration yet.