Discrimination against Pre-PhD Students

Our target group is contacted, and we measure early-career everyday discrimination towards potential future researchers.

2026-06-02

2026-06-16

Response rate in the survey.

Dummy variable which is equal to one if the researcher has participated in the survey, (i.e. clicked on the survey link in the email / opened link in the browser), and zero otherwise.
For this variable, it does not matter which or how many answers have been provided in the survey.

Effort Provision
We will repeat the analyses for the primary outcome variable for this secondary outcome.

Number of answers provided in the survey.
We count the provision of consent as the first answer. If a researcher actively chooses the option to not provide consent, their “effort provision” will be equal to one. Subjects can answer 7 questions (including consent provision) in Part 1 of the survey and 110 questions in Part 2. Hence, the maximum number of answers is 117. The number of answers can be zero even if the researcher has participated in the survey (by clicking on the link but not answering the question on consent). If a researcher participates multiple times in the survey, we will consider participation with the maximum number of answers as the one relevant to measure their effort provision.

See experimental design (hidden)

Not available

Randomization is done in python using stochatreat. We stratify the list of researchers into four strata:
1) Gender (Female/Male),
2) Field (Economics/Business Administration),
3) Country (Germany, Switzerland, Austria), and
4) Type of Position (Tenured/Non-Tenured).

Randomization is done in 2 steps.

1) All researchers, excluding those who are employed at the University of Cologne, University of Mannheim or Ruhr-University Bochum, will be randomized into 36 groups as described above.
2) For each of the excluded universities in step 1 separately, we randomly allocate the researchers into 24 groups as described above.
3) We combine the groups of steps 1 and 2 to have 36 groups in total. Each student will contact researchers of two groups, with none of the groups containing researchers from their own university. To one of their groups, they will send emails including the expression of interest in doing a PhD. To the other group, they will send emails without that addition.

Receiver of the email

No

Treatment is not clustered

The number of observations corresponds to the number of professors and postdoctoral researchers whose emails we have collected. Our final sample contains 4472 researchers and their email addresses.

We will randomly send each researcher one email which is i) either sent by a man or a woman and ii) indicates interest in pursuing a PhD or not.

9 male students and 9 female students will send the emails. Our sample contains 1375 female and 3097 male researchers.

See below for the sample sizes in the treatment arms:

Male Sender:
- PhD Interest Mail & Male Researcher: 769 researchers/ planned emails
- Mail w/o PhD Interest & Male Researcher: 776 researchers/planned emails
- PhD Interest Mail & Female Researcher: 347 researchers/planned emails
- Mail w/o PhD Interest & Female Researcher: 333 researchers/planned emails

Female Sender:
- PhD Interest Mail & Male Researcher: 783 researchers/planned emails
- Mail w/o PhD Interest & Male Researcher: 769 researchers/planned emails
- PhD Interest Mail & Female Researcher: 343 researchers/planned emails
- Mail w/o PhD Interest & Female Researcher: 352 researchers/planned emails