|
Field
Trial Title
|
Before
Can competition undermine taste-based ethnic discrimination during the hiring process?
|
After
Can output competition in output performance undermine ethnic discrimination during the hiring process?
|
|
Field
Abstract
|
Before
To examine if the competition in the labour market can reduce taste-based ethnic discrimination during the hiring process, we will use a lab experiment to stimulate the hiring process. Participants will be asked to finish several real effort tasks in the experiment, to capture the differences in actual ability(aka productivity) and the willingness to compete between the ethnic majority group and the minority group. The hiring process includes three treatments. In the baseline treatment, all the subjects are asked to choose a partner from the rest without any restriction. In the second treatment, all the subjects will face an actual employer competition for the “popular candidates”, who have been selected by more than one subject. The third treatment, some subjects will be given additional information about the choice of other “employers”, to find the differences between discriminatory behaviour and strategic reactions. To minimise statistical discrimination, language skill and the actual ability will be implicitly informed during the hiring process.
|
After
To examine if the competition in the labour market can reduce taste-based ethnic discrimination during the hiring process, we will use a lab experiment to stimulate the hiring process. In this study, we are interested in the effect of output competition on ethnic discrimination during the hiring process. The main experiment is a hiring game with 2x2 treatments. In each session, participants will play a role as employers and they are required to hire one of the candidates as their employees and form a group. The 2x2 treatments differ by 1) whether the ethic information of employee candidates is given 2) whether they have an output competition with the other groups. Employers payments will depend on their employee's performance and the rankings under the treatments with output competition. Through this experiment, we expect to capture the effects of output competition on taste-based ethnic discrimination.
|
|
Field
Trial End Date
|
Before
November 23, 2022
|
After
February 28, 2024
|
|
Field
Last Published
|
Before
June 23, 2021 02:23 AM
|
After
October 27, 2021 11:12 PM
|
|
Field
Intervention (Public)
|
Before
1) Baseline: no employer competition
2) Treatment - employer competition: All the subjects will face an actual employer competition for those who have been selected by more than one subject. The following task is randomly assigned to be competitive or non-competitive (stimulating the competition in the long-term performance)
3) Treatment - Strategic interactions: Some subjects can receive additional information about the choice of other employers, during the hiring process. This is to distinguish the taste-based discrimination from the strategic interaction.
|
After
1) Baseline (A): No ethnic information is given in profiles, no output competition
2) Baseline- employer competition (B): No ethnic information is given in profiles, output competition exists
3) Baseline - Type (C): Ethnic information is given in profiles, no output competition
4) Treatment (D): Ethnic information is given in profiles, output competition exists.
|
|
Field
Intervention Start Date
|
Before
December 01, 2020
|
After
December 01, 2021
|
|
Field
Intervention End Date
|
Before
August 31, 2021
|
After
February 28, 2024
|
|
Field
Primary Outcomes (End Points)
|
Before
For each individual employee, i, whether i is being selcted by at least one employer from the majority group.
|
After
Probability of a minority candidate being hired, controlled by given scores (rank), age, and treatments.
|
|
Field
Primary Outcomes (Explanation)
|
Before
ethnical-sounding names (determined by reported birthplace), previous relative performance of i, size of the sessions, competitive hiring market (=1), competitive performance (=1), updating (=1)
|
After
1) Expect no difference in hiring decision between Treatment A and Treatment B
2) The probability of being hired is lower in Treatment C than in Treatment D
3) The probability of a minority being hired are lower in Treatment C than in Treatment A
4) Expect no difference in hiring decision between Treatment A or Treatment B, compared to Treatment D
|
|
Field
Experimental Design (Public)
|
Before
The experiment includes two parts. The first part is to ask the subjects to finish several real effort tasks, to capture their differences in ability (aka productivity) and in the willingness to compete. The second part is to stimulate the recruitment process in the hiring market, with three different treatment. Treatment 1 is the baseline that all the subjects are allowed to select a partner freely to finish another real effort task. Treatment 2 is to introduce the employer competition in the selection process that all the subjects will face an actual employer competition for the popular candidates, who have been selected as 1st preference by more than one subject. Treatment 3 is to update the information about the choice of other "employers" to some subjects, and see how they respond in changing their hiring decision. The first two treatments are to test the competition effect on taste-based discrimination. And the last two treatments are to distinguish the non-profitable taste-based discrimination from profitable strategic interactions. The statistical discrimination is minimised by explicitly informing the actual productivity and language level during the hiring process. The in-group variable is determined by the reported birthplace. To avoid bias in self-creating identity, some subjects are randomly chosen as anonymous during the hiring process.
|
After
The experiment includes two parts. The first part is the preliminary phase and the second part is a hiring game. In the preliminary phase, we will ask participants to finish an individual experiment, including 1) an individual task repeated five times 2) an exit survey. And at each time, participants have 2 minutes to finish.
We will classify all the profiles into two groups 1) ethnic majority group and 2) ethnic minority group.
The second part consists of a hiring game. We will recruit participants from Prolific, who only satisfy the majority identification. In the hiring game, all the employers have to choose and hire 1 employee based on given profiles of individuals recruited from the preliminary phase to finish the game. Each session has 6 profiles from the majority group, and 6 profiles from the minority group. We have 4 different treatments with differences in 1) given information in the profiles of employees candidates and 2) Competition in output performance.
|
|
Field
Planned Number of Clusters
|
Before
120-200 subjects
|
After
Each session has 2 participants. 100 sessions for each treatment. The total will be 800 individuals.
|
|
Field
Planned Number of Observations
|
Before
120-200 observations
|
After
Each session has 12 profiles. Total will be 4800 observations.
|
|
Field
Sample size (or number of clusters) by treatment arms
|
Before
at least 20 sessions (5-8 subjects per session)
|
After
100 sessions
|
|
Field
Intervention (Hidden)
|
Before
Real effort task: 4 letter word anagram tasks because
According to the previous lab study in racial discrimination (Wozniak & MacNeill, 2020), the ability between white and black in 4-6 letter word anagram is indifferent. Therefore, we can assume the ability between the natives and minority group 1 (native English speakers) should be indifferent.
The difficulty of anagram-solving task should decrease as the number of letter decrease (Nioka et al., 2008). In order to reduce the ability differences across different groups, I choose 4 letter rather than 3 letter is considering there more words in 4 letters. So the given questions are less likely to repeat in different tasks.
Word anagram task requires a certain level of English skills, so it is reasonable to assume a pervasive stereotype against non-native English speakers, which will generate implicit discrimination against the minority groups (Reuben et al., 2014). And therefore we would observe the change in discriminatory behaviour clearly via the experiment.
Hypothesis:
1) The ethical discrimination against the minorities (both native and non-native English speakers) should decrease in the competitive hiring market, compared to non-competitive hiring market, if the hiring strategies of other firms are not allowed to observe or respond (without updating).
2) The ethical discrimination against the minorities (both native and non-native English speakers) should increase in the competitive hiring market, compared to non-competitive hiring market, if the hiring strategies of other firms are not allowed to observe or respond (with updating).
3) IF the ethnical discrimination against the minorities are higher in the competitive hiring market with or without strategic interactions, it might be due to the statistical discrimination against minorities because of
- A belief that difference in language skill will result in differences in ability (testing if the minority employers select differently between a oversea native English speaker vs. an oversea non-native English speaker.
- A belief that minorities have lower willingness to compete and therefore will perform worse than the majorities (testing if the minority employers select differently between a local English speaker (majority) and an oversea English speaker.
|
After
The participants of the hiring game are recruited from the ethnic majority groups.
In the first two treatments (Treatment A and Treatment B, the profiles of employee candidates do not include ``type" information to the employers during the selection process, i.e., which ethnic group that each employee candidate belongs to. In both treatments, profiles also include candidates' scores of a task, which is randomly drawn from the five tasks completed before by the candidates during the preliminary phase. This is to minimise statistical discrimination and also introduce a noisy signal for hiring decisions. The total number of years of higher education completed by each employee will also be presented in the profiles to reduce the priming effect of ``type" information.
The difference between Treatment A and Treatment B is the competition in output performance. In Treatment A, employers have no competition in the output: they will receive the payment based on the scores of another randomly drawn task of their selected employee, and they will earn £0.25 per score. In Treatment B, employers have competition in the output: only 2 employers will get paid based on the scores of another randomly drawn task of their selected employee. An employer that choose the top 1 best performer will receive £0.8 pounds per score while employer that choose the top 2 best performers can receive £0.4 pound per score.
The other two treatments (Treatment C and Treatment D) will clearly inform the "type" information to the employers in the profiles during the selection process. The differences between Treatment C (no output competition) and D (with output competition) are exactly the same as the differences between Treatment A and B.
|
|
Field
Secondary Outcomes (End Points)
|
Before
For each individual employer, I, whether they choose select a candidate of same ethnic group as 1st preference
|
After
|
|
Field
Secondary Outcomes (Explanation)
|
Before
ethnical-sounding names (determined by reported birthplace), previous relative performance of 1st preference, size of the sessions, competitive hiring market (=1), competitive performance (=1), updating (=1)
|
After
|
|
Field
Building on Existing Work
|
Before
|
After
Yes
|