Principal Investigators: Melina Platas Izama, Pia Raffler
Country: Uganda

Registration: 20150820AA

Dates of Intervention: May 2015 – February 2016 (Primary elections Summer 2015 ; General elections February 2016)

Background: What is the effect of information on political behavior? This field experiment, conducted in Uganda during the 2015 primary and 2016 general elections, will systematically assess the conditions under which information about candidates and government performance affects voter behavior. We examine two different methods of providing information: debate-like “Meet the Candidate” sessions and a scorecard. “Meet the Candidate” sessions include video-recorded candidate statements on a set of questions related to policy preferences. These sessions will be publicly screened in one set of polling stations and privately to individuals in another set of polling stations. The screenings will take place in both an intra-party and inter-party electoral environment, in the 2015 primary elections of the ruling party, and 2016 general elections. Thus, we examine systematically two factors that we hypothesize will affect the effect of information on voter behavior: the political environment and the public vs. private nature of information provision.

Research Design: The common treatment arm makes use of scorecards: local government and service delivery audit information will be hand delivered to individuals during surveys. The alternate treatment arm consists of public and private screenings of “Meet the Candidate” sessions and surveys in primary and general elections, and assesses the role of public information in affecting voter attitudes and behavior. Conducting two types of screenings – public and private – will illuminate the role of public information in affecting voter attitudes and behavior.

Hypotheses:
  • Political information (exposure to screenings) will increase a) political knowledge, b) voter turnout, c) votes for the candidate who is perceived to perform best in the “Meet the Candidate” session.
  • Political information (exposure to debates) will have a larger effect on votes for the winner, voter turnout, and political knowledge in an intra-party than an inter-party environment.
  • A candidate who performs poorly in the “Meet the Candidate” session, (compared to viewer expectations) will reduce turnout in an inter-party environment to a greater extent
    than in an intra-party environment. In a highly polarized environment, supporters of the lower performing candidate in the session will switch their vote to another candidate
    in an intra-party environment, but will not in an inter-party environment and will be less likely to vote at all in the case of the latter.
  • The effect of political information provided in a group setting will have a greater effect on voting behavior than information provided in an individual setting. That is, support (measured as intended vote choice) for the perceived best performer of the session will be greater in the group than individual screening.
  • There will be lower support for the incumbent or the incumbent’s party in constituencies where the results of the scorecard/audit place the constituency below the median constituency performance, while support for the incumbent/incumbent’s party will be higher where the results of the scorecard/audit place the constituency above the median constituency performance.