Effects of group size on investment decisions & Inconsistent Risk Preferences

Last registered on April 29, 2026
View Trial History

Pre-Trial

Trial Information

General Information

Title
Effects of group size on investment decisions & Inconsistent Risk Preferences
RCT ID
AEARCTR-0018505
Initial registration date
April 29, 2026

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published
April 29, 2026, 4:29 PM EDT

First published corresponds to when the trial was first made public on the Registry after being reviewed.

Locations

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

Request Information

Primary Investigator

Name
Sascha Füllbrunn
Affiliation
Radboud University
Contact Primary Investigator

Other Primary Investigator(s)

PI Name
Wolfgang Luhan
PI Affiliation
Portsmouth University
Contact Investigator

Additional Trial Information

Status
In development
Start date
2026-05-10
End date
2026-11-10
Keywords
Behavior, Finance & Microfinance, Other
Additional Keywords
Experiment, Decision Making For Others, Principal-Agent, Risk Preferences
JEL code(s)
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
Abstract
The study consists of two distinct experiments that each participant will take part in.

Experiment 1 (Responsibility): This project studies how the size of the affected group influences decisions made on behalf of others. We implement a controlled online experiment in which the number of principals varies systematically while convex incentive structures are kept constant across conditions. This design isolates the effect of group size from payoff differences. Participants make decisions under risk for others across a wide range of group sizes, including large-scale settings. The approach enables a clean identification of how group size shapes responsibility and risk-taking.

Experiment 2 (Inconsistency): This study tests whether risk preferences elicited via the Gneezy–Potters investment task are consistent across different expected value conditions. In a within-subjects design, participants make investment decisions under three scenarios that vary only in expected returns while keeping the overall structure constant. The experiment randomizes condition order and pays out only one condition to limit hedging behavior. By comparing decisions across conditions, the design evaluates whether standard classifications of risk preferences remain stable. The results provide a direct assessment of the reliability of commonly used experimental risk measures.
External Link(s)

Registration Citation

Citation
Füllbrunn, Sascha and Wolfgang Luhan. 2026. "Effects of group size on investment decisions & Inconsistent Risk Preferences." AEA RCT Registry. April 29. https://doi.org/10.1257/rct.18505-1.0
Sponsors & Partners
Experimental Details

Interventions

Intervention(s)
For experiment 1, we will vary the group size when making risky decisions for other participants using a within-subjects design. Between-subjects, we vary the order in which the group size will be presented.
For experiment 2, we will vary the expected value of a risky asset using a Gneezy-Potters (1997) experimental design within subjects. Between subjects, we vary the order in which the different treatments are presented.
Intervention Start Date
2026-05-10
Intervention End Date
2026-11-10

Primary Outcomes

Primary Outcomes (end points)
Experiment 1: Principals have an endowment and the decision maker decides on how much to invest. The amount invested is the key outcome variable.

Compared to earlier experiments on Prolific, we want to find out whether the laboratory environment reduces the social distance and therefore leads to a stronger group size effect, compared to a strong order effect of group sizes found in Prolific experiments.

Primary Outcomes (explanation)
none will be constructed

Secondary Outcomes

Secondary Outcomes (end points)
Experiment 2: The decision maker has an endowment to invest. The amount invested in the three environments POS, ZERO, and NEG are the three outcome variable.

We have the following hypotheses:
1. Incentive Compatibility: Investment increases with expected value (Inv(POS)>INV(ZERO)>INV(NEG)
2. Stability of Risk Preferences: Risk-preference classifications inferred in the POS environment remain stable across environments. That is, if a participant is classified as risk-averse, risk-neutral, or risk-seeking in POS, the same classification should apply in ZERO and NEG.
3. Rank-Order Persistence: If individual risk-preference classifications are not stable across environments, the relative ordering of participants’ investment levels should still persist across POS, ZERO, and NEG.
Secondary Outcomes (explanation)
none will be constructed

Experimental Design

Experimental Design
Experiment 1: 101 participants will make risky decisions for other participants of varying group size using Qualtrics.
Experiment 2: 101 participants will make risky decisions for themselves with varying expected returns using Qualtrics.
Experimental Design Details
Not available
Randomization Method
The software programme Qualtrics will randomly allocate the participants to the treatments (1-5-10-20-50-100 and 100-50-20-10-5-1, and pos-zero-neg and neg-zero-pos in almost equal share.
Randomization Unit
Student participant. They will use the same survey over all sessions such that the programme applies the randomisation to each session
Was the treatment clustered?
No

Experiment Characteristics

Sample size: planned number of clusters
No clusters
Sample size: planned number of observations
101 participants (in the extreme case, 1 decision maker decides for 100 other participants)
Sample size (or number of clusters) by treatment arms
There are always 50/51 participant in each order treatment but 101 participants for the within-subject design consideration.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
Accounting for the within-subject design, the unit of analysis is the participant, not the individual decision. For a regression, investment_it=alpha_i+beta x log(GroupSize_t) + epsilon_it with GroupSize_t = {1,5,10,20,50,100}, 101 participants, 80% power and alpha = 0.05, the detectable standardized within-subject effect is d_z =[ z_0.975+z_0.2 ]/square root(101) = (1.96+0.84)/10.05 = 0.28 SD. This corresponds to roughly 6–8 percentage points under typical assumptions for bounded investment data (20-30 percentage point SD of investments), and up to 14 percentage points under a conservative upper bound (50 percentage points SD of investments). When assuming a similar regression with pos = 1, zero = 0, and neg = -1 in place of log(GroupSize_t), having a directed hypothesis (pos>zero>neg), we get d_z =[ z_0.95+z_0.2 ]/square root(101) = (1.645+0.84)/10.05 = 0.25 SD with the same interpretation.
Supporting Documents and Materials
IRB

Institutional Review Boards (IRBs)

IRB Name
Faculty Ethics Committee, of Faculty of Business & Law, University of Portmouth
IRB Approval Date
2026-04-27
IRB Approval Number
BAL/2026/19/LUHAN
Analysis Plan