Social Marginal Welfare Weights

Last registered on February 09, 2024


Trial Information

General Information

Social Marginal Welfare Weights
Initial registration date
November 29, 2021

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
December 03, 2021, 1:48 PM EST

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

Last updated
February 09, 2024, 8:14 AM EST

Last updated is the most recent time when changes to the trial's registration were published.



Primary Investigator

University of Zurich

Other Primary Investigator(s)

PI Affiliation
Erasmus University Rotterdam

Additional Trial Information

Start date
End date
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
In this project, we estimate ``Generalized Social Marginal Welfare Weights" (henceforth welfare weights), introduced by Saez and Stantcheva (2016), using an experimental approach. Welfare weights measure the value of providing an additional dollar of consumption to any individual in the society. These weights are general enough to capture many different fairness ideals. In a variety of policy settings, such as taxation, cash transfers, and in-kind transfers, welfare weights can be used to compare alternative policies. To understand how people assign welfare weights, we administer a large-scale incentivized online experiment in a general population sample of the U.S. population. In our experiment, participants in the role of ``Social Architects," assign welfare weights to seven real ``Recipients" with different real-world after-tax incomes. We provide the first estimate of welfare weights using a general population sample of the U.S. population. The weights obtained from our project can be directly used to evaluate several policies such as taxation, cash transfers, and in-kind transfers. We shed light on the factors (both demographics and political affiliations) that affect people's welfare preferences. Finally, we compare the welfare weights obtained from our experiment to the weights implied by existing policies in the U.S.
External Link(s)

Registration Citation

Capozza, Francesco and Krishna Srinivasan. 2024. "Social Marginal Welfare Weights." AEA RCT Registry. February 09.
Sponsors & Partners

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

Request Information
Experimental Details


Intervention Start Date
Intervention End Date

Primary Outcomes

Primary Outcomes (end points)
Slope of the weights
Weight assigned by an Architect to each Recipient
Primary Outcomes (explanation)
Slope of the weights: this is the slope coefficient obtained by regressing an Architect's weights on the vector (-1,-2,-3,-4,-5,-6,-7)

Secondary Outcomes

Secondary Outcomes (end points)
Government should do something to reduce inequality (1-7)
Government should increase top-taxes (1-7)
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
We recruit the participants in the role of Social Architects or Recipients. In the main task, Social Architects face 6 decision screens. In each decision screen, they face a different pair of Recipients (Recipient i and Recipient j) and have to decide how to allocate some money between them. A Social Architect faces a ``staircase" with 4 questions within each Decision Screen. In each question, the Social Architect has to indicate whether she prefers a Constant reform or a Variable reform. The Constant reform option takes $500 away from Recipient j and gives $500 to Recipient i. The variable Reform option takes away $t from Recipient j and gives $pt to Recipient i.

The four questions that Social Architects are selected from a set of 15 questions based on the staircase procedure. The first question that Social Architects face is common for all Social Architects. The second, third, and fourth questions that Social Architects face depend on the choices that the Social Architects made in the first, second, and third questions respectively. In each decision screen, we take the Social Architects' choices in the fourth question to identify the relative weight (the value of p) they assign to the two Recipients. The selected value of p is the value that makes the Social Architect just indifferent between a Constant reform (500,-500) and Variable Reform (pt,-t).

After the task of assigning weights, Social Architects face a second task where we elicit their policy views. The first question asks them whether they would like to increase the tax on millionaires and the second question asks them if they would like the government to increase redistribution.

We inform the Social Architects that there is a chance that they will be randomly selected in this study. At the end of the study, we will randomly select one Social Architect. For the randomly selected Social Architect, one of the six decision screens will be randomly selected, and one of the four questions within the selected decision screen will be randomly selected and implemented. At the end of the study, the randomly selected question will involve two Recipients. We will recruit these two Recipients from a survey panel. The bonus payments of the two Recipients will depend on the choices of the randomly selected Social Architect.

The experimental flow described above is for the Loss x Moderate treatment in the study. We implement other treatments to assess how robust the welfare weights estimation is. In treatment Gain x Moderate, we change the framing of the reforms such that the reform amounts are all positive. In treatments Loss x Moderate and Gain x Moderate, every Recipient is compared to the Recipient earning $70000. Treatments Loss x High and Gain x High are similar to treatments Loss x Moderate and Gain x Moderate, with the exception that all the Recipients are compared to the Recipient earning $500,000.
Experimental Design Details
Randomization Method
Qualtrics random number generator
Randomization Unit
Was the treatment clustered?

Experiment Characteristics

Sample size: planned number of clusters
Sample size: planned number of observations
2000 participants in Wave 1 of data collection
Sample size (or number of clusters) by treatment arms
500 participants per treatment
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
1. Slope of the weights - Mean = 0.02, sd = 0.08, n per group = 500, power = 80%, minimum detectable effect size = 0.014 (70% increase over mean). 2. Weights assigned to a Recipient - N/A

Institutional Review Boards (IRBs)

IRB Name
The Human Subjects Committee of the Faculty of Economics, Business Administration and Information Technology at the University of Zurich
IRB Approval Date
IRB Approval Number
IRB Name
Internal Review Board, section experimental research, of the Erasmus School of Economics, Erasmus University Rotterdam
IRB Approval Date
IRB Approval Number
Analysis Plan

Analysis Plan Documents

Pre-analysis plan: 14 December 2022 - v2

MD5: 267f4ac69a5087db7590453235365ec6

SHA1: 31a8be53dba4df687aae16d7c38682dd7679f6e1

Uploaded At: December 14, 2022


Post Trial Information

Study Withdrawal

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

Request Information


Is the intervention completed?
Data Collection Complete
Data Publication

Data Publication

Is public data available?

Program Files

Program Files
Reports, Papers & Other Materials

Relevant Paper(s)

Reports & Other Materials