"There is now abundant data that diversity is essential to scientific excellence, with experiments showing that diverse teams have cognitive diversity that allows them to outperform homogeneous teams." (Columbia University 2018)
Unfortunately, research shows that people tend to form their networks with people who are similar to them, which can hinder diversity in fields that have traditionally been shaped by networks of white men, such as academia.
How diverse are German research projects regarding the gender ratio of their teams?
How gender diverse are individual researchers' collaborations?
Let's find out from data by one of the largest German institutions to fund research projects: Deutsche Forschungsgemeinschaft.
They do not provide information on the gender of their research, so with a little help, we had to infer a gender based on the researcher's first name.
This is the gender distribution of all researchers funded by the DFG.
To stay critical of this data and validate our inference, we used a subset of the data that contains only professors.
Since the title "Professor" / "Professorin" has a gendered connotation in German, we could compare the gendered title to our inferred gender.
Please also note that we were unable to infer outside of the gender binary due to the lack of such a dataset.
Mistakes are represented in yellow, red and blue and contain three error classes: a mismatch between title and inferred gender, a gender could not be assigned or the first name was not recognized.
This is the ratio of correct gender inference for the subset of professors.
Looking at each funded project with more than one person, we can now calculate the gender ratio per project.
Thus, the project gender index indicates the gender ratio in a project, where 0 means there are only men and 1 means there are only women.
In detail, 10488 (67%) collaborative projects funded by the DFG involve only men, whereas 667 (4%) involve only women.
This histogram shows the distribution of the project gender index.
You can now look at individual institutions and their gender index, based on the researchers associated with them.
The dot size is based on the amount of projects an institution's researchers are involved in.
This map shows the institutions and their top 10 researchers individually.
So, how and where can we start addressing the fact that research collaborations are clearly skewed towards men-only groups? Let's start in places with the most impact.
Based on the amount of projects that an institution's researchers are involved in, these are the top 30 institutions.