Gender discrimination in hiring processes in Bogota: A correspondence Study

This intervention aims to determine whether employers that post online vacancies discriminate against men and women applicants based on their potential and effective parental situation. We conduct a correspondence study in one of Colombia’s largest online platforms for job seekers, where we apply with six different profiles that vary by sex, effective parental status, and potential parental status. We will apply to different types of sectors and occupations to try and capture differences across economic activities that are male or female dominated. We expect this evidence to help disentangle the effect of discrimination due to potential and effective parenthood in online job applications, for which there is scarce evidence, especially in a developing country setting.

External Link(s)

Registration Citation

Citation

Ham, Andres et al. 2024. "Gender discrimination in hiring processes in Bogota: A correspondence Study." AEA RCT Registry. January 19. https://doi.org/10.1257/rct.12416-1.0

Sponsors & Partners

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

Request Information

Experimental Details

Interventions

Intervention(s)

We want to explore whether employers discriminate against workers based on potential/effective parenthood and sex. Hence, we will send 6 resumes and profiles to each vacancy in an online job repository as follows:

i) female + potential mother + children sign
ii) female + potential mother + no children sign
iii) female + not potential mother + children sign
iv) female + not potential mother + no children sign
v) Male + children sign
vi) Male + no children sign

Intervention Start Date

2023-10-16

Intervention End Date

2023-12-15

Primary Outcomes

Primary Outcomes (end points)

Callback rates

Primary Outcomes (explanation)

Our main outcome variable is a binary variable that is zero if the profile does not receive a callback for an interview and one if the profile receives a callback

Secondary Outcomes

Secondary Outcomes (end points)

Secondary Outcomes (explanation)

Experimental Design

We will send 6 resumes and profiles to each vacancy combined as follows:

i) female + potential mother + children sign
ii) female + potential mother + no children sign
iii) female + not potential mother + children sign
iv) female + not potential mother + no children sign
v) Male + children sign
vi) Male + no children sign

This design will allow us to i) differentiate discrimination based on potential and effective parenthood, and ii) study discrimination patterns between and within sex groups. In the case of effective parenthood, we will signal the candidate's parental status by including or not including the father/mother's membership in a school's parent association in the resume. Furthermore, in the case of potential parenthood, we will signal the candidate's potential to have children in the future through the age. The age of potential mothers will randomly vary between 20 and 30 years old, and the age of non-potential mothers will randomly vary between 40 and 50 years old. For male applicants, the age will be the same of the female applicants.

Experimental Design Details

Randomization Method

Each vacancy we apply to will receive applications from all six profiles

Randomization Unit

Vacancy (job posting)

Was the treatment clustered?

Yes

Experiment Characteristics

Sample size: planned number of clusters

1000 job postings (vacancies)

Sample size: planned number of observations

6,000 applications to 1,000 job postings (vacancies)

Sample size (or number of clusters) by treatment arms

2,000 applications control (male); 2,000 applications T1 (female potential mothers), 2,000 applications T2 (female effective mothers)

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

We conduct power calculations by simulation since we have no pre-intervention data on callback rates for individuals. We consider different number of vacancies where we will apply with the six profiles. We simulate a binary outcome with a normal distribution with mean 0.05 (5% callback rate) and standard deviation 0.25, calculating power for different effect sizes (from 1 to 10 percentage points). This process is repeated 10,000 times for each fixed number of vacancies and effect size. Sending six applications to each vacancy would allow us to estimate a minimum detectable effect of 0.032 or 3.2 percentage points from the control group mean

Supporting Documents and Materials

IRB