Social secrecy and voter behavior

How to cite

Agneman, Gustav. 2022. "Social secrecy and voter behavior." AEA RCT Registry. AEARCTR-0009066. https://doi.org/10.1257/rct.9066-1.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Gustav Agneman*

* Primary investigator

Registry ID
AEARCTR-0009066
DOI
https://doi.org/10.1257/rct.9066-1.0
First registered
Last updated
Status
On going
Trial dates
2022-01-22 – 2022-03-21
Countries
Colombia (Meta)
JEL codes
D72, D91
Registry record
http://www.socialscienceregistry.org/trials/9066

Abstract

Social secrecy refers to voters’ beliefs as to whether they will reveal their vote choice to anyone (be it friends, family, or strangers). Despite the fact that social secrecy theoretically alters the costs associated with certain vote choices, few studies have directed attention toward how social secrecy (or a lack thereof) influences voter behavior. In this study, I conduct an RCT that aims to induce induce experimental variation in social secrecy. I then leverage this variation how social secrecy impacts voter behavior.

Intervention

The RCT consists of a baseline survey that measures, among other things, the social secrecy among respondents. I randomly divide respondents into three groups: one group receives two text messages that remind respondents that they can keep their vote secret; another group receives three texts that remind them that they can keep their votes secret and that doing so means that their vote is completely anonymous; a third group which does not receive any texts and thus functions as the control group.

The timing of the intervention is as follows:

March 1. Both group 1 and group 2 receive a text that reads: “For the upcoming parliamentary elections, we would like to remind you that you have the right to cast your vote in private. You do not have to share your voting choices with anyone.”

March 4. Group 2 receives a text that reads: “If you do not tell anyone which party you will vote for or have voted for, no one can find out.”

March 9. Both group 1 and group 2 receive a text that reads: “Even if someone asks you, you do not need to reveal who you vote for. Your vote is your business.”

Experimental design

The study builds on a simple combination of a baseline survey, intervention, and an endline survey. Parliamentary elections are held in-between the intervention and the endline survey. Accordingly, the questions on voting behavior in the endline survey will regard voter behavior during a real election. The endline survey will be conducted between March 14th and March 21st.

Design details

Randomization method
I use the Randomizr package in R to assign respondents into treatment groups.
Randomization unit
The treatments were individual and respondents were assigned into groups using block-randomization (which ensured that treatment and control groups were equally large within each municipality).
Clusters
No clusters (individual randomizations).
Observations
The number of respondents is expected to be in the range of 600-650.
Arms
The treatment arms are roughly equally large (~ 230 in each arm). The response rate for the endline survey is not going to be 100 percent and, hence, the final number of observations will be lower.
Primary outcomes
The primary outcomes of interests are social secrecy, institutional secrecy, (perceived) voter secrecy, and voter choice (both the actual vote choice and the motivation, captured through the following questions: "When deciding how to vote, to what extent do you take into account what other people might think about your vote choice?"; "Please describe in 1-3 sentences why you chose to vote the way you did.").
Secondary outcomes
Voter participation could also potentially be impacted by social secrecy. Another outcome of potential interest is whether respondents experienced vote-buying attempts and, if so, how they dealt with it. I measure this both using List Experiments and a direct survey question.

Cited by

No citation data available for this registration yet.

Keywords