Motivated statistical discrimination

Last registered on November 30, 2021
View Trial History

Pre-Trial

Trial Information

General Information

Title
Motivated statistical discrimination
RCT ID
AEARCTR-0007991
Initial registration date
August 02, 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
August 02, 2021, 5:31 PM EDT

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

Last updated
November 30, 2021, 2:24 PM EST

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

Locations

Country
United States of America
Region

Primary Investigator

Name
Markus Eyting
Affiliation
JGU Mainz
Contact Primary Investigator

Other Primary Investigator(s)

Additional Trial Information

Status
In development
Start date
2021-12-01
End date
2021-12-31
Keywords
Behavior
Additional Keywords
Discrimination, Information
JEL code(s)
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
Abstract
Researchers typically categorize discrimination as either taste-based or belief-based discrimination based on either accurate or inaccurate beliefs. I argue, that when an agent does not hold randomly inaccurate beliefs, but instead holds inaccurate beliefs due to motivated reasoning, resulting discrimination looks like inaccurate statistical discrimination when it really is a form of motivated discrimination. While any kind of discrimination can have fatal consequences for the discriminated, a precise identification of the source of discrimination has important implications for effective policy interventions. In order to have a closer look at discrimination based on inaccurate beliefs I suggest a careful analysis of the formation of these inaccuracies. In particular analyze the effect of biased information processing on discriminatory behavior and show that people selectively attend to and interpret information in line with their motives - if they have the necessary 'wiggle room' to do so.
External Link(s)

Registration Citation

Citation
Eyting, Markus. 2021. "Motivated statistical discrimination." AEA RCT Registry. November 30. https://doi.org/10.1257/rct.7991-3.0
Sponsors & Partners
Experimental Details

Interventions

Intervention(s)
Intervention Start Date
2021-12-01
Intervention End Date
2021-12-31

Primary Outcomes

Primary Outcomes (end points)
"Discrimination", "information acquisition" and "information processing"
Primary Outcomes (explanation)
"Discrimination": I measure the frequencies with which workers of different subgroups are 'hired' and to what extent hiring is consistent with the seen signals in each treatment.
"Information acquisition": I measure how many signals employers gather in each treatment and whether or not specific signal requests depend on the interaction of the treatment group and the previous signal(s).
"Information processing": I measure to what extent hiring decisions are consistent with previously seen information signals between treatments.

I analyze whether results differ when looking at decisions between blacks and whites vs. between whites and hispanics vs. between asians vs. hispanics. This means, I look at whether discrimination within minority groups (asians vs. hispanics) is different than discrimination between a majority and a minority group (whites vs. blacks; whites vs. hispanics).

(I check and account for learning and fatigue effects by dropping observations in which decision times significantly deviate from the median.)

Secondary Outcomes

Secondary Outcomes (end points)
Decision times
Beliefs
Secondary Outcomes (explanation)
"Decision times": I measure the time taken to make a hiring decision.
"Beliefs": I measure the subjective probabilities regarding the scores of particular worker subgroups (blacks, whites, hispanics, asians) before and after an aggregate information update

Experimental Design

Experimental Design
I set up a hiring situation in which 'employers' are repeatedly asked to select one of two potential 'workers'.

Employers are randomly allocated to the groups "neutral" and "real". In group 'neutral' any information about whether or not the workers belong to a minority group is hidden.

In each hiring decision, employers first receive a randomly drawn piece of information about the two workers. They may then choose to gather additional information signals before hiring one of the two workers.
Experimental Design Details
The pool of workers has been set up via an online survey in which US students completed an assessment test consisting of a real effort task, a dictator game and a logic/math quiz.

Previous to the hiring stage, employers in both groups are provided with the true group statistics about the average productivity of different
subgroups of workers (in order to correct and align prior beliefs between groups real and neutral). All employers are asked to state their beliefs about the subgroups' scores before and after having seen the true group statistics.

In the "neutral" group any information about whether or not the worker belongs to a minority group is hidden. Instead, workers are given abstract group memberships. In the "real" group, employers are presented with the races of the two workers.

In both groups, employers are then repeatedly presented with two profiles of workers (subjects from the worker pool) and asked to 'hire' one of them. Employers are paid a fixed fee if they hired the worker who has received the higher score in a randomly drawn decision. In both groups, the employers are initially given one random piece of information about the two workers (e.g. their current college GPA) and have the opportunity to gather more information signals about the two workers.
Randomization Method
Randomization into groups "neutral" and "real" is done by a computer
Randomization Unit
Individual
Was the treatment clustered?
No

Experiment Characteristics

Sample size: planned number of clusters
800 individuals
Sample size: planned number of observations
~10000 decisions
Sample size (or number of clusters) by treatment arms
250 individuals in group real, 250 individuals in group neutral
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
Supporting Documents and Materials
IRB

Institutional Review Boards (IRBs)

IRB Name
Gemeinsame Ethikkommission Wirtschaftswissenschaften der Goethe-Universität Frankfurt und der Johannes Gutenberg-Universität Mainz
IRB Approval Date
2021-03-23
IRB Approval Number
N/A
Analysis Plan

Post-Trial

Post Trial Information

Study Withdrawal

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

Request Information

Intervention

Is the intervention completed?
No
Data Collection Complete
Data Publication

Data Publication

Is public data available?
No

Program Files

Program Files
Reports, Papers & Other Materials

Relevant Paper(s)

Reports & Other Materials