NEW UPDATE: Completed trials may now upload and register supplementary documents (e.g. null results reports, populated pre-analysis plans, or post-trial results reports) in the Post Trial section under Reports, Papers, & Other Materials.
Watching the State: Can New Technologies Promote (a Sense of) Democracy?
Initial registration date
August 29, 2019
September 04, 2019 7:55 PM EDT
University of California San Diego
Other Primary Investigator(s)
Additional Trial Information
Why do autocrats implement new technologies aimed at increasing transparency - to curb corruption or to create a veneer of democracy? This project studies the effects of technology of video monitoring of election on its outcomes, the behavior of local authorities and citizens' beliefs in one prominent authoritarian regime, Russia. Using administrative data on results of the 2018 Presidential election, I show that video monitoring suppressed voter turnout by 5% and reduced the number of votes cast for the incumbent by 8.5%, consistent with an improvement in electoral accountability. However, these effects were partially mitigated by displacement of votes to nearby unmonitored polling stations. To test the mechanisms and evaluate the effects of video monitoring on citizens’ beliefs and behavior, I conduct online and in-person survey experiments ahead of the 2019 Single Voting Day. In particular, I remind respondents in the treatment group that many polling stations in the upcoming election will have video monitoring that anyone can freely stream on the government-run website. Afterward, I elicit respondents' voting behavior and beliefs, perceptions about elections, and general views on democracy.
The intervention consists of two survey experiments conducted online and in-person. The treatment group receives a reminder that many polling stations in the upcoming election will be equipped with video monitoring. It also provides information that the government-run website <name> will freely stream the election day online. The control group does not receive any information about video monitoring. After the treatment is introduced, I elicit respondents’ voting behavior and beliefs, trust in elections, and general views on democracy. To elicit voter intimidation, I cross-randomize a listing experiment over the main treatment.
Intervention Start Date
Intervention End Date
Primary Outcomes (end points)
Voting behavior, Voter intimidation, Trust in elections, Views on democracy
Primary Outcomes (explanation)
Voting Behavior: An index that combines outcomes of self-reported turnout, voting for the ruling party, and second-order beliefs about the voting behavior of others.
Voter Intimidation: The number of items named in an embedded listing experiment that is cross-randomized over the main experiment.
Trust in Elections: An index that combines outcomes of trust in election results and beliefs about elections leading to improvements.
Views on Democracy: An index that combines outcomes of a state of democracy in Russia, perceptions of democratic values, and beliefs on tax compliance.
Secondary Outcomes (end points)
Knowledge about video monitoring, Engagement in video monitoring, Effectiveness of video monitoring
Secondary Outcomes (explanation)
In-person survey. About two weeks before Single Voting Day, I will conduct a survey experiment on a subsample of respondents of a nationally representative survey who indicate that there will be elections in their locality. Half of this subsample will be assigned to a treatment group which will be reminded about the presence of video monitoring at many polling stations. They will also be given information that the government-led website <name> will stream the election day online and asked whether they will watch it. Respondents in the control group will not receive any information about video monitoring.
Online survey. Wave 1. Two weeks before Single Voting Day, I will use the online survey platform Anketolog to recruit 1,600 respondents who reside in the regions that will have Gubernatorial elections. Half of them (800 respondents) will be assigned to a treatment group that will get a reminder about video monitoring in the upcoming elections. Another half (800 respondents) will not get any information about video monitoring. Afterward, I will elicit respondents’ intended voting behavior, beliefs about the voting behavior of others, and perceptions of electoral integrity.
Wave 2. After Single Voting Day, I will attempt to recontact the same sample of 1,600 respondents. I will elicit their real voting behavior, perceptions of electoral integrity, and general views on democracy. I will also cross-randomize half of them into a listing experiment to elicit voter intimidation. Moreover, I will elicit their knowledge and engagement in video monitoring, as well as their beliefs on its effectiveness.
Experimental Design Details
Treatment Questions (unique for both surveys):
(1) Did you know that many polling stations in the upcoming election will also be equipped with webcams that will stream the election day online?
(2) The website <name> will stream the election day online from all polling stations in the country that are equipped with webcams. You can simply visit the <name> on the election day and observe the voting at any polling station. Will you observe the election on <name>?
Randomization is conducted by survey firms.
Randomization is done at individual level.
Was the treatment clustered?
Sample size: planned number of clusters
Treatment is not clustered. Information on the number of observations is presented below.
Sample size: planned number of observations
(1) In-person survey: A subsample of a nationally representative sample of 1,600 respondents who will indicate that there will be elections in their locality on the 2019 Single Voting Day (approximately 60% of the full sample according to the electoral calendar).
(2) Online survey: 1600 respondents who reside in the regions that will have gubernatorial elections on the 2019 Single Voting Day.
Sample size (or number of clusters) by treatment arms
In-person survey: 1/4 of the subsample that has elections control-no listing, 1/4 of the subsample control-listing, 1/4 of the subsample treatment-no listing, 1/4 of the subsample treatment-listing.
Online survey: 1/4 of the sample (~400 respondents) control-no listing, 1/4 of the sample (~400 respondents) control-listing, 1/4 of the sample (~400 respondents) treatment-no listing, 1/4 of the sample (~400 respondents) treatment-listing. Note: There is a probability of attrition from the 2nd wave of the online survey.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
INSTITUTIONAL REVIEW BOARDS (IRBs)
UCSD Human Research Protections Program (HRPP)
IRB Approval Date
IRB Approval Number
Post Trial Information
Is the intervention completed?
Is data collection complete?