Payment Schemes and Fairness Experiment 3

Eliciting fairness ratings for different payment schemes, eliciting willingness to pay for implementing a specific payment mechanism, varying how much participants are affected by their own decision

2024-09-18

2024-11-30

(i) Fairness ratings (overall, outcome, procedural) of payment mechanisms
(ii) Willingness to pay for implementing a payment mechanism

(i) Participants rate the overall, outcome, and procedural fairness of two different payment mechanisms
(ii) Participants can pay part of their show-up fee to implement the piece rate mechanism (which was perceived as fairer in past studies) in a laboratory experiment. They are informed that their decision is only implemented, and money deducted from their show-up fee, if they are randomly chosen to determine the payment mechanism of the lab experiment.

(i) Demographics (age, gender, political orientation, assessment of life satisfaction)

(i) Answers will be used as control variables. Assessment of life satisfaction uses Cantril ladder

Main experiment is conducted in an online session with two randomized treatments. Additionally, we conduct two lab sessions in which participants will play the task described in the online survey with one of the payment mechanisms chosen by random participants of the online survey.

Our study consists of two experiments. The main experiment is conducted in an online survey with 300 participants from the pool of the AWI lab in Heidelberg, who earn a fixed amount of €6. Participants are presented with a real-effort slider task from a previous study. Subjects in this previous study were organized in groups of two and tried to correctly place as many sliders correctly as possible in the given time frame. The payment for this task depended on one of 16 possible payment mechanisms. The task was repeated four times, with subjects getting additional or less time in subsequent rounds in some of the payment mechanisms.

Participants in the current experiment read the description of this task and are then presented with two of the possible payment mechanisms. The first mechanism is a tournament, where whoever placed more sliders correctly receives a large price of €3, while the loser gets 0.3€. Additionally, the winner got 6 seconds more in the next round, the loser 6 seconds less. This tournament thus exacerbates differences between the two participants. The second mechanism is a piece rate, where participants are renumerated based on the number of sliders each of them places correctly, with no further time rewards or penalties.

After reading a description of each of these mechanisms, participants answer three sets of questions. The first set of questions directly asked on the same page as the mechanism description are comprehension questions. Participants are confronted with a hypothetical scenario of two players playing the task in the first round. They then need to state the correct earnings of both players, as well as their time in seconds for the next round. The next set of questions consists only of one question eliciting the participants’ overall fairness perception of the mechanism on a 0 to 10 Likert scale. The third and final set of questions asks the participant to rate the procedural and outcome fairness of the mechanism, once again on a 0 to 10 Likert scale. When answering this questions, participants can take a look at the task description and the description of the payment mechanism.

Following reading and rating the two payment mechanisms, participants learn that we will play the described slider task in a laboratory session. They are also informed that they can give up part of their fixed payment to implement the piece rate mechanism rather than the tournament. This willingness to forgo payment is elicited using a multiple price list with 11 steps (see instructions for details). Each step implies forgoing €0.3 of the fixed payment to implement the piece rate mechanism. They are also reminded of their overall fairness rating of both mechanisms on this page. Participants are told that one participant from the survey will be picked at random. If they are picked, their choice will be implemented, and they will receive the potentially reduced fixed payment. If they are not picked, their choice will not be implemented, and they receive the fixed payment of €6 (thus even if they would be willing to forgo a large part were their decision to be implemented).
We have two treatments that will become relevant at this stage. Participants in the impartial treatment will encounter the situation exactly described as above. Hence, they know that others will play the slider task in a laboratory session and that their choice will potentially affect them. This lab session will be played by 20 participants recruited from the AWI lab pool. Participants in the partial treatment on the other hand are told that they will receive an invitation to another lab session of 20 people. Hence, they know that their choice can potentially affect themselves.

Randomization of treatments done by the experimental software

Individual

No

320 Participants

320 Participants

150 participants in impartial treatment (online)
150 participants in partial treatment (online, 20 of which will be invited to the partial lab session)
20 participants in the impartial lab session

We assume a significance level of 5% and a power level of 80% in the following. First, participants in the first two experiments we conducted rated the overall fairness of the piece rate as 5.375 out of 10. The overall fairness rating of the tournament was 8.485. In order to identify the same difference within person (using a paired-means t-test), we would need 23 participants, even when assuming a fairly high common standard deviation of 5. Our main test will be a test of the willingness to forgo part of the fixed payment to potentially implement the piece rate mechanism. Assuming that participants in the impartial treatment are not willing to forgo any part of their payment and participants in the partial treatment are willing to give up €0.3 (the smallest possible amount in the price list). In that case, we would identify this difference as significant (using a two-sample t-test) in our sample of 300, as long as the standard deviation is not above 0.92 (hence more than three times the difference). Assuming the same standard deviation, if participants were willing to forgo 0.6€ (two steps on the multiple price list), we would only need 38 participants per treatment to significantly identify this effect.