Experimental Design Details
We will recruit two samples: one sample consisting of university students, and one sample consisting of professionals. Our student-sample will be recruited from two different universities where students participate in computer-based experiments in the universities’ respective laboratories. The student sample will be split evenly across universities. Both the treatment and control group will contain students from both universities. We split the sample in order to minimize the influence of university-specific student characteristics that may influence the results.
Our professional sample will be collected using the database of a professional sample provider. We purchase access to their database to be able to directly target finance professionals from the field. These professionals will take part in the study via an online platform.
Participants will receive a performance-based financial compensation. The participants’ payoffs will depend on three randomly selected choices and will be adjusted with a predetermined exchange rate to ensure that payoffs equal the hourly wage of a student research assistant, on average.
The participants will face a series of decision problems inspired by Tversky and Kahneman (1992). Student participants will face 56 variants of the following decision problems. Professional participants will face fewer variants due to time constraints. Participants choose between a risky lottery on the one hand (e.g., 25% chance to win 150€ and a 75% chance to win 50€) and a certain outcome on the other hand (e.g., 100% chance to win ##€). For each prospect, seven certain outcomes are shown. These are logarithmically spaced between the lowest and highest outcome of the risky lottery. For each of the certain outcomes, the participant chooses between either the risky lottery or the certain outcome.
In a second step, participants will refine their choices by again deciding between the risky lottery and seven certain outcomes, which now are equally spaced between a value slightly lower than lowest certain outcome accepted in the first set and a value slightly higher than the highest certain outcome rejected.
For each decision problem, internal consistency of choices will be checked after the participants have submitted their choices. If a participant chooses inconsistently within a decision problem, they will be facing the same decision problem once again with the indication that the decisions have been inconsistent. While all participants will be facing the same decision problems, the order will be randomized.
We will further survey all participants for basic demographic information and personality traits:
- gender, age, educational degree, study major, financial expertise,
- extraversion, agreeableness, need for affiliation, social comparison tendencies
In more detail, we will use the following scales:
Evaluation of personality characteristics (on scales of 1-5, 1 being “strongly disagree”, 5 being “strongly agree”):
- extraversion, agreeableness: Big Five Inventory (12 items each Soto & John, 2017)
- Need for affiliation: Steers & Braunstein, 1976 (5 items)
- social comparison: INCOM Scale (Gibbons & Buunk, 1999; 6 items)
Participants will be randomly assigned to either a control group or to our treatment.
1) In the control group, participants will face the task as described before without having any interaction with peers. In essence, the control group is a replication of the original work of Tversky and Kahneman (1992).
2) In our treatment, participants will receive information on the payoffs of randomly assigned peers. More precisely, participants will first decide on 50% of the lotteries, assigned at random. For the remaining 50% of the lotteries, participants will learn about the outcome of (partly) randomly matched peers for the respective lottery. That is, for each lottery in the second half (50%) of the lotteries, participants will be presented the payoff that one particular peer received on this lottery. Before participants decide on the first decision problem with peer feedback, they will be assigned and introduced to two peers. For each following lottery they will see the outcome of one of those two peers. In effect, we make use of a stratified sampling technique and ensure that participants observe only payoffs from peers who have shown a consistent pattern throughout their lottery decision problems. Thereby, we de facto create three treatment groups: i) one condition with, on average, more risk averse peers (than the participant before the treatment occurs), ii) one condition with, on average, equally risk averse peers (compared to the participant before the treatment occurs), and iii) one condition with, on average, less risk averse peers (than the participant before the treatment occurs).
We conduct an exit survey using three different pages. The survey contains the following items.
- Do you think the researchers in this study had an agenda? (yes or no)
- If yes, please state what do you think the research agenda was.
What did the authors try to show with this study? (multiple choices allowed)
- The authors tried to show how people make monetary choices.
- The authors tried to show that people evaluate probabilities non-rationally.
- The authors tried to show that people are risk-seeking.
- The authors tried to show that people are influenced by the results of others.
- The authors tried to show that people can make many decisions adequately in a short time
- Please give the names of your two peers (treatment conditions only).
- Do you think your peers were either risk-averse or risk-seeking? (5-point scale; 1 being ”risk-seeking” to 5 being “risk-averse”) (treatment conditions only)
- While playing the lotteries, did you feel closely connected to your peers? (5-point scale; 1 being “strongly disagree” to 5 being “strongly agree”)
- Did a comparison with your peers’ outcomes influence your choices? (5-point scale; 1 being ``strongly disagree'' to 5 being “strongly agree”)
- The comparison with my peers’ outcomes made me take more risky choices than I usually would have? (5-point scale; 1 being “strongly disagree” to 5 being “strongly agree”)
- The comparison with my peers’ outcomes made me take less risk than I usually would have? (5-point scale; 1 being “strongly disagree” to 5 being “strongly agree”)
Anonymity measure: Participants’ study data will be separately stored from their personal information that is needed to handle the performance-based financial compensation.
Miscellaneous: All experiments will be implemented using oTree (Chen et al, 2016).