Disclosure Discrimination

Last registered on December 06, 2021

View Trial History

Pre-Trial

Trial Information

General Information

Title

Disclosure Discrimination

RCT ID

AEARCTR-0008662

Initial registration date

December 04, 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 06, 2021, 10:22 AM EST

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

Locations

Country

Czechia

Region

Primary Investigator

Name

Darya Korlyakova

Affiliation

CERGE-EI, a joint workplace of Charles University and the Economics Institute of the Czech Academy of Sciences, Politických vězňů 7, 111 21 Prague, Czech Republic

Contact Primary Investigator

Other Primary Investigator(s)

PI Name

Rastislav Rehák

PI Affiliation

CERGE-EI, a joint workplace of Charles University and the Economics Institute of the Czech Academy of Sciences

Contact Investigator

PI Name

Sona Badalyan

PI Affiliation

CERGE-EI, a joint workplace of Charles University and the Economics Institute of the Czech Academy of Sciences

Contact Investigator

Additional Trial Information

Status

In development

Start date

2021-12-05

End date

2022-03-01

Keywords

Gender, Labor, Other

Additional Keywords

Information, Disclosure, Hiring, Discrimination, Immigrants, Women, Online Experiment

JEL code(s)

C90, D83, J71

Secondary IDs

Prior work

This trial does not extend or rely on any prior RCTs.

Abstract

In this project, we aim to explore the existence of labor market discrimination in a novel context - communication among hiring team members. Widely used correspondence studies focus on discrimination in callbacks or interview invitations, an early stage of the hiring process. However, for instance, Quillian, Lee, and Oliver (2020) document substantial additional discrimination from interview to job offer. Based on discussion in Dobbin and Kalev (2016) and our interviews with professionals from recruitment industry, we hypothesize that human resource assistants, who are often responsible for sharing their notes from interviewing candidates with (hiring) managers, consciously or unconsciously disclose or emphasize different information about candidates depending on candidates’ ethnicity or gender. To test this conjecture, we conduct an online experiment with a representative sample of Czech respondents whom we ask to perform HR assistants’ job. Each assistant sees multiple profiles of workers and can select any information from them. In each profile, we vary exogenously and independently worker’s nationality and gender by manipulating the name. Assistants know that the selected information is intended to be shown to other respondents whom we assign the role of hiring managers. Collecting data from this complementary sample is worthwhile because it will enable us to study the consequences of assistants’ decisions for hiring of females and immigrants at later stages.

Sources
Dobbin, F., & Kalev, A. (2016). Why diversity programs fail. Harvard Business Review, 94(7), 14.
Quillian, L., Lee, J. J., & Oliver, M. (2020). Evidence from field experiments in hiring shows substantial additional racial discrimination after the callback. Social Forces, 99(2), 732-759.

External Link(s)

Registration Citation

Citation

Badalyan, Sona, Darya Korlyakova and Rastislav Rehák. 2021. "Disclosure Discrimination." AEA RCT Registry. December 06. https://doi.org/10.1257/rct.8662-1.0

Sponsors & Partners

Experimental Details

Interventions

Intervention(s)

Intervention (Hidden)

Intervention Start Date

2021-12-05

Intervention End Date

2022-03-01

Primary Outcomes

Primary Outcomes (end points)

Number of disclosed information pieces and share of disclosed information pieces (overall and per profile sections). Profiles include the following sections: (personal) summary, basic information (e.g. demographics), education, work, certificates, skills, interests, and volunteering (in some profiles).

We plan to explore heterogeneity along the following dimensions: profile quality, assistants’ gender and attitudes towards women and foreigners.

Primary Outcomes (explanation)

Secondary Outcomes

Secondary Outcomes (end points)

1) To gain further insights into disclosure discrimination, we plan to focus on the effects of worker’s gender on particular information pieces (especially age, marital status, number of kids, and parental leave). We also plan to focus on the effects of worker’s gender and nationality on the features of disclosed information – valence and relevance. To group information pieces objectively along these two criteria, we plan to recruit another sample of respondents (similar to our experimental participants).

2) Time spent by assistant on a worker’s profile, number of clicks on learn-more buttons.

We plan to explore heterogeneity along the following dimensions: profile quality, assistants’ gender and attitudes towards women and foreigners.

Secondary Outcomes (explanation)

Experimental Design

Each experimental subject (assistant) will see 8 workers' profiles (one at a time) and for each of them he/she will be asked to select pieces of information for another randomly matched respondent (manager). The manager will decide how much he or she is willing to pay to hire each of the 8 workers for a financial task.

We have constructed 8 profiles based on actual data from a supplementary survey with workers. We include both high- and low-quality profiles. Profiles contain various information about workers' demographics, his/her education, work experience, certificates (that demonstrate workers' results on real-effort tasks), self-reported skills, etc. Each of the 8 profiles has 4 versions which have identical content but differ in workers’ fictitious names. The exogenous manipulation of names aims to signal gender and nationality. We use 2x2 factorial design and assign workers' profiles to female or male version (dimension 1) and to local (Czech) or immigrant version (dimension 2). We focus on immigrants from former Soviet states and use Russian and Ukrainian names.

Assistants will be incentivized to select information carefully because managers could reward their effort based on the quality of disclosed information. We will also make assistants aware that their decisions may have financial consequences for other respondents. Particularly, hired workers and managers (see below) can receive bonuses.

After the experiment with assistants is completed, we will conduct a supplementary survey with managers. Managers will be responsible for hiring workers for a financial task. More precisely, they will set wages and, if the chosen wage for a particular worker is higher than a randomly generated number, the worker will be hired. One of the 8 hiring decisions will be chosen at random and implemented: the manager will receive a bonus equal to the difference between the worker’s performance on a financial task and the random number and the worker will receive the random number as a bonus. Before making each of the 8 hiring decisions, the manager will see information about the worker from a random assistant. While pairing assistants and managers, we will use fixed matching and assign a different assistant to each manager.

Experimental Design Details

Randomization Method

Randomization through Qualtrics

Randomization Unit

For every assistant and each of the 8 profiles that the assistant sees, we will assign randomly and orthogonally gender and nationality to the profile, independently of every other profile and every other assistant.

Was the treatment clustered?

Experiment Characteristics

Sample size: planned number of clusters

Sample size: planned number of observations

6000 observations (750 assistants x 8 workers’ profiles)

Sample size (or number of clusters) by treatment arms

about 1500 observations

Minimum detectable effect size for main outcomes (accounting for sample design and clustering)

This sample size is sufficient to detect the treatment effect of size 0.0143 in terms of share of disclosed information pieces with 80% power and 5% significance level. This sample size is sufficient to detect the treatment effect of size 0.445 in terms of number of disclosed information pieces with 80% power and 5% significance level.

Supporting Documents and Materials

IRB