|
Field
Abstract
|
Before
We plan to experimentally test the effect of voting rules on the likelihood that individuals vote for moral transgression. Subjects vote independently from each other but monetary benefits are equally divided among group members. Our first hypothesis is based on a theoretical model by Rothenhäusler et al. (2018) which (in case of consequentialist moral costs) shows that the number of votes for moral transgression increases in the number of votes required (voting threshold). Our second hypothesis is that guilt sharing among group members is a main driver of this result.
|
After
We plan to experimentally test the effect of voting rules on the likelihood that individuals vote for moral transgression. Subjects vote independently from each other but monetary benefits are equally divided among group members. Our first hypothesis is based on a theoretical model which (in case of consequentialist moral costs) shows that the number of votes for moral transgression increases in the number of votes required (voting threshold). Our second hypothesis is that guilt sharing among group members is a main driver of this result. We test the second hypotheses by a sequential elimination of other motivational factors (financial incentives, up-dates on the preferences of group members and social conformity).
|
|
Field
Trial End Date
|
Before
December 31, 2019
|
After
March 31, 2020
|
|
Field
Last Published
|
Before
July 24, 2019 02:49 PM
|
After
December 28, 2019 03:06 PM
|
|
Field
Intervention End Date
|
Before
December 31, 2019
|
After
March 31, 2020
|
|
Field
Primary Outcomes (End Points)
|
Before
Our primary outcome is whether subjects vote for or against taking money originally designated for donation to a charity depending on the treatment (voting threshold).
|
After
Our primary outcome is whether subjects vote for or against taking money originally designated for donation to a charity depending on the setting (voting threshold) and different treatments.
|
|
Field
Experimental Design (Public)
|
Before
The experiment will be conducted online and participants will be recruited using Amazon Mechanical Turk.
Subjects will be randomly assigned to groups of three. They vote on whether to take money designated for donation to a charity for their group (“Yes” or “No”). If, depending on the voting threshold, sufficiently many group members vote in favor of taking the money, the payoff is split equally among group members.
We test our hypotheses with different treatments (one for each possible voting threshold) and assign subjects to exactly one of these treatments (between-subject design):
T1: At least one “Yes” vote is required for transferring the money to the group. If all group members vote “No”, the money is donated.
T2: At least two “Yes” votes are required for transferring the money to the group. If at least two group members vote “No”, the money is donated.
T3: All group members need to vote “Yes” for transferring the money to the group. If at least one group members votes “No”, the money is donated.
In each of these three treatments, two out of the three group members make their vote unconditional, i.e. independent of the other group members' votes. These observations will be used to analyze the first hypothesis (on the impact of the voting threshold).
The third group member’s vote will be elicited conditional on the other group members’ votes. As this vote matters only for the group outcome in case the third vote is pivotal, we only ask for the following scenarios depending on the treatment:
T1: What would you vote for if none of your two group members voted “Yes”?
T2: What would you vote for if exactly one of your two group members voted “Yes”?
T3: What would you vote for if both of your two group members voted “Yes”?
Comparing the votes of the third group members who know that their vote is pivotal for the three thresholds allows us to test the second hypothesis (on the impact of guilt sharing). The reason is that the marginal (financial) benefit of voting “Yes” is the same in each treatment.
|
After
The experiment will be conducted online and participants will be recruited using Amazon Mechanical Turk.
Subjects will be randomly assigned to groups of three. They vote on whether to take money designated for donation to a charity for their group (“Yes” or “No”). If, depending on the voting threshold, sufficiently many group members vote in favor of taking the money, the payoff is split equally among group members.
We test our hypotheses with different settings (one for each possible voting threshold) and assign subjects to exactly one of these settings (between-subject design):
Threshold 1: At least one “Yes” vote is required for transferring the money to the group. If all group members vote “No”, the money is donated.
Threshold 2: At least two “Yes” votes are required for transferring the money to the group. If at least two group members vote “No”, the money is donated.
Threshold 3: All group members need to vote “Yes” for transferring the money to the group. If at least one group member votes “No”, the money is donated.
In the first treatment, we are going to ask participants to vote unconditionally, i.e. independent of the other group members’ votes. These observations will be used to analyze the first hypothesis (on the impact of the voting threshold).
In the second treatment, we are going to split the group. Two randomly selected voters will decide unconditionally. The third group member will be asked for their vote conditional on the other two group members’ votes. The questions will be:
What would you vote for if none of your two group members voted “Yes”?
What would you vote for if exactly one of your two group members voted “Yes”?
What would you vote for if both of your two group members voted “Yes”?
Comparing the votes of the third group members who know that their vote is pivotal for the three thresholds allows us to test the second hypothesis (on the impact of guilt sharing). The reason is that the marginal (financial) benefit of voting “Yes” is the same in each treatment.
However, even when financial incentives are the same due to pivotality, there may still be different motivational factors including e.g. social conformity. To disentangle these factors, we will have two additional treatments. In these treatments, the actual votes of the group members are substituted by the votes of other participants.
In the third treatment, conditional voters are informed about the behavior of unconditional voters from the second treatment and the behavior of their group members. However, the votes of their group members are substituted with the votes of two random conditional voters from the second treatment.
In the fourth treatment, conditional voters are in the same situation as in the third treatment. However, they receive no information about the behavior of their actual group members.
Furthermore, we are also going to replicate the second treatment with a fifth treatment where we inform conditional voters about the behavior of unconditional voters from the second treatment.
|
|
Field
Planned Number of Observations
|
Before
We plan to collect 1080 observations, i.e. 1080 participants.
|
After
We plan to collect 4680 observations, i.e. 4680 participants.
|
|
Field
Sample size (or number of clusters) by treatment arms
|
Before
We plan to collect 360 observations per treatment, i.e. 360 participants in each treatment.
|
After
We plan to collect 360 observations in the first treatment and 1080 observations in each of the other treatments.
|
|
Field
Power calculation: Minimum Detectable Effect Size for Main Outcomes
|
Before
In a previous experiment with a similar general setup, we found that in a lying task 14% of participants lied in T1, while 27% (28%) lied in T3 (T2). Assuming that we can replicate this effect in our donation experiment, we need about 120 observations (one-sided Chi-squared test with significance level of 5% and power of 80%) for conditional votes per treatment. As conditional votes account for one third of observations per treatment, we need 360 observations per treatment and 1080 observations in total.
|
After
In a previous experiment with a similar general setup, we found that in a lying task 14% of participants lied in “Threshold 1”, while 27% (28%) lied in “Threshold 3” (“Threshold 2”). Assuming that we can replicate this effect in our donation experiment, we need about 120 observations (one-sided Chi-squared test with significance level of 5% and power of 80%) for each setting per treatment. As conditional votes account for one third of observations in all but the first treatment, we need 1080 observations per treatment (in treatments 2 to 5) and 4680 observations in total.
|