Please fill out this short user survey of only 3 questions in order to help us improve the site. We appreciate your feedback!
The Influence of Public Participation on Voter Turnout
Last registered on May 07, 2021


Trial Information
General Information
The Influence of Public Participation on Voter Turnout
Initial registration date
May 07, 2021
Last updated
May 07, 2021 10:33 AM EDT

This section is unavailable to the public. Use the button below to request access to this information.

Request Information
Primary Investigator
National Taiwan University
Other Primary Investigator(s)
PI Affiliation
National Taiwan University
PI Affiliation
National Taiwan University
Additional Trial Information
On going
Start date
End date
Secondary IDs
More evidence shows that public movement can promote voting turnout in the following election. This paper wishes to go further with these findings to answer how people’s experience of public participation affects their voting behavior. We develop a two-stage lab experiment to test what factors in public participation mainly change people’s voting behavior. In the first stage, subjects participate in threshold public good games in different settings including low threshold, communication and opinion leader. Then, in the second stage, every subject has to decide whether to vote. We can thus observe people’s voting behavior with various participation experiences. Our initial result shows that low threshold is easier to reach than the baseline and thus subjects in this treatment have more experiences of success. However, such successful experience does not promote voting rate. Communication and opinion leader also creates more successful experience for subjects and effectively promote voting turnout. We believe communication and opinion leader are factors that mainly influence people’s voting behavior, while pure successful experiences have nearly no effect.
External Link(s)
Registration Citation
Chen, Josie I, Ting-Wei Chen and Chun-Fang Chiang. 2021. "The Influence of Public Participation on Voter Turnout." AEA RCT Registry. May 07. https://doi.org/10.1257/rct.7622-1.0.
Experimental Details
Intervention Start Date
Intervention End Date
Primary Outcomes
Primary Outcomes (end points)
1. Individual voting behavior (the subjects choose to vote or abstain)
2. Success (the number of rounds in which each group’s total contribution exceeds the threshold)
3. Group contribution (the number of total contributions made by group members in each round)
Primary Outcomes (explanation)
Secondary Outcomes
Secondary Outcomes (end points)

Secondary Outcomes (explanation)
Experimental Design
Experimental Design
Subjects have to undergo a two-stage experiment. In the first stage, each subjects except members of the control group is assigned to a group of 10 members and plays multiple rounds of threshold public good games under different settings. In the second stage, subjects will play a one-shot voting game. We plan to have one control group and four or five treatment groups (baseline, low-threshold, communication, opinion leader and re-shuffle). The re-shuffle treatment will be executed only if the results from the control group significantly differ from the baseline group.
In the baseline group, subjects play threshold public good games with threshold being 40 tokens. Subjects can decide to contribute any amount out of their endowment which is 10 tokens. They will win a prize of 15 tokens if their total contribution exceeds the threshold.
Leader and Communication are key factors that facilitate public participation. To understand the effects of communication with and without opinion leader, we have one communication group and one leader group, and subjects in these two groups are allowed to communicate in the first stage. The threshold remains the same as the threshold of baseline. In the communication group, subjects can communicate with other group members for 2 minutes to discuss strategies before the first round. The channel of communication is cheap talk and the content is unrestricted except personal information.
In the opinion leader group, an opinion leader is endogenously decided by a personality test. Subjects have to respectively choose five adjectives that best describe their own personalities and five adjectives best describe a leader’s personalities from 42 options. The selection method is similar to Borda Count. The score of each personality is the number of group members who regard each personality as leader’s personality. Then, we sum up the total score of the five personalities which best describe subjects’ own personalities. The subject with the highest score is selected as the opinion leader. An opinion leader has 60 seconds to send one-way messages to the group members in the beginning. The content of the messages is unrestricted except personal information. After leader sends the messages, group members have a 2-minute chat which is the same as that in the communication treatment group.
Public participation with leader or communication might directly affects voting behavior, but it might affect voting behavior through other channels like experience of success. On the one hand, successful experiences may promote voter turnout in the second stage because of the mutual trust was established in the first stage and people are conditional cooperators. On the other hand, successful experiences in the first stage may reduce subjects’ willingness to vote due to free-ride motives. To investigate the effects of successful experiences in the first stage on the voter turnout in the second stage, we have a low-threshold group in which the threshold in the first stage is reduced from 40 tokens to 20 tokens to increase the probability of passing the threshold in the first stage. By comparing the voting turnout in low-threshold and baseline groups in the second stage, we will have the effects of experiencing success.
In the second stage, all subjects in Baseline, Leader, Communication, Opinion Leader groups play the same voting game. Each subject is assigned to a 10-person group and the group members remain the same as the previous stage. Subject has an endowment of 50 tokens. Subjects have to pay the voting cost of 35 tokens if they decide to vote. Subjects win a prize of 100 tokens if their vote is more than the rivalry group, while they win nothing if they lose the voting game.
We design a “control game” for the control group in which subjects will not have teammates from the lab at the same time in the first stage. In the first stage, subjects in control group also have to decide their contribution to an account in the control game. However, their contributions will be matched with historical data of the baseline group. Subjects can win a prize of 15 tokens if subject’s own contribution adding the value of total contribution in the historical data exceeds the threshold of 40. To avoid subject share cooperating experience with the historical subjects, subjects are told that the value matching their contribution is randomly drawn in an unknown distribution. In the second stage, they will be assigned to a group to play the same voting game as subjects in other groups. In sum, the subjects in the control group are deprived of the experience of public participation. In this design, the difference between control and baseline may imply the effect of the participating experience in the threshold public good game.
The experience of participation actually contains various effects. As the public good stage is replaced, the group identity formulated in that stage also disappears. We cannot solely specify if group identity affects the voting rate by the design of control. If the difference is little, there is no need to go further to investigate the effect of group identity. On the other hand, if the difference is large enough, we have to run another treatment group called “Re-shuffle” to clarify such effect. In the re-shuffle group, the setting in the first stage is completely the same as the baseline. However, in the second stage, the rule of deciding payoff is slightly different from the baseline in the re-shuffle group. In this treatment, one of the group members is randomly drawn to match the historical data. This subject’s vote adding the historical voting turnout comparing with the rivalry group decides if the group members can get the payoff.
Experimental Design Details
Not available
Randomization Method
Randomization done in Laboratory by a computer
Randomization Unit
Was the treatment clustered?
Experiment Characteristics
Sample size: planned number of clusters
48 groups (10 individuals in each group)

Sample size: planned number of observations
480 individuals
Sample size (or number of clusters) by treatment arms
There are six treatment arms: control, baseline, low-threshold, communication, opinion leader and reshuffle
8 groups for the control
8 groups for the baseline
8 groups for the low-threshold
8 groups for the communication
8 groups for the opinion leader
8 groups for the reshuffle
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
IRB Name
Research Ethics Committee, National Taiwan University
IRB Approval Date
IRB Approval Number
IRB Name
Research Ethics Committee, National Taiwan University
IRB Approval Date
IRB Approval Number
Analysis Plan

There are documents in this trial unavailable to the public. Use the button below to request access to this information.

Request Information