Responding to Positive and Negative Past Interactions_Online

How to cite

Wicker, Till. 2025. "Responding to Positive and Negative Past Interactions_Online." AEA RCT Registry. AEARCTR-0016047. https://doi.org/10.1257/rct.16047-1.1

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Till Wicker*

* Primary investigator

Registry ID
AEARCTR-0016047
DOI
https://doi.org/10.1257/rct.16047-1.1
First registered
Last updated
Status
In development
Trial dates
2025-05-23 – 2025-08-11
Countries
Private
JEL codes
D91, J71
Registry record
http://www.socialscienceregistry.org/trials/16047

Abstract

Discrimination is widespread across the world, and a variety of contexts. Discrimination has further been widely studied through the use of field experiments (for an overview, see Bertrand and Duflo, 2017), with two dominant theories emerging: taste-based (Becker, 1957) and statistical discrimination (Phelps, 1972; Arrow, 1973). However, the study of discrimination thus far in the literature has been static: studies have looked at one-off interactions and the response to discrimination in those. Past work in Uganda (IRB FUL 2024-015) has illustrated that discrimination among refugees and Ugandans is dynamic in the sense that past interactions influence discriminatory behavior in the future. In particular, perceived negative discrimination in the past increased the likelihood of negative discrimination in the future. This is not the case with positive discrimination. The aim of this project is to document the persistence of discrimination among a different sub-population, and identify the mechanisms more cleanly through different sub-treatments. I have since developed a theoretical model to rationalize these results, which creates novel predictions that this new project will test.

Intervention

The study will run over two weeks, with two parts. the first part mirrors the lab-in-the-field experiment in Uganda, where across two stages, a participant is both a worker completing tasks, and a manager delegating tasks. In this case, the focus will be on ethnic discrimination in the US among men. Managers in the first stage will be either white or black (signaled by their name), and will either allocate tasks favorably, equally, or unfavorably for the participant. After completing the task, the participant becomes the manager in the next stage, and allocates tasks between two workers: one white and one black. In this setting, I introduce a third stage, where participants will have 1 minute time to solve as many tasks as possible. The name of the manager in this stage is disclosed to the participants. This is to test for micro-foundations of anticipated discrimination.

One week later, participants are invited back for a second round of the game. First, they are asked to recall how many tasks they were assigned last week. Afterwards, they play the game again, this time as a manager.
Subsequently, they are shown 10 allocations across managers and workers, and are incentivized to recall the names of the managers and allocations accordingly. Subsequently, they play the game again, as a manager. Afterwards, they will be assigned one of the rounds of the game, as a worker. Then, then will be randomized across 6 different arms: 1) the normal game; 2) direct retaliation against the previous manager; 3) costly mistakes; 4) inefficient non-even allocation; 5) letting players know there are future rounds; 6) non-white, non-black player.
Afterwards, there will be a PEQ, asking about discriminatory attitudes, and the recall of negative and positive past experiences with managers.
With a different sample, the same setup will be used, however to look at the same setup among white men (with an affirmative action policy component); as well as to look at the role of previous joyful destruction, a dictator game setup, and noisy signals on future discriminatory behavior
Following a pilot study of the previous iteration, the focus has been placed on racial discrimination, rather than gender-based discrimination.
The current experimental design allows to distinguish between four micro-foundations of retaliatory discrimination, which have been pre-registered with this registry.
50% of respondents will fill in detailed beliefs in the first stage of the experiment.

Experimental design

A lab experiment will be conducted on Prolific. Individuals will be randomized across several treatments, as outlined in more detail above.
The allocation of tasks is pre-determined, based on data from a pilot study.

Design details

Randomization method
Randomization is done via qualtrics
Randomization unit
individual
Clusters
Same as the number of observations
Observations
100 observations per treatment arm.
Arms
100 observations per treatment arm. Fewer for the sub-samples
Primary outcomes
Allocation of tasks across workers in second stage. Time taken to complete task in first stage. Recall of past events shown. Effort put in into third effort task.

Cited by

No citation data available for this registration yet.

Keywords