How diverse is my ‘science team’? Looking at Colombian coauthorships in OpenAlex

How diverse is our team’s cognitive spectrum? From how many countries are we? How many men or women are in the group? Is a man or woman leading or driving the project?

This is a third piece on teams in science, the first one on a definition and a second one on diversity, here and here

Team diversity in Colombia. Source: www.openalex.org publications with institutional affiliation from Colombia between 2014–2023. Documents with 11 or more coauthors were excluded. Note: Team leadership — There are 3 types i) ‘female lead’ where first-last authors are women, ii) ‘male lead’ where first-last authors are men, and iii) ‘gender paired’ where first-last authors are man-woman or woman-man. Team composition — There are 3 types i) ‘female driven’ is a team where over a half of members are women, ii) ‘male driven’ where over half of members are men, and iii) if they are 50/50 it is ‘gender paired’. X-axis: number of unique research field ‘expertise’ of coauthors per research document (e.g., if 3 there are coauthors who research fields expertise are in three different areas: biotech, nanotech, and chemical engineering). Y-axis: team composition + number of unique countries of institutional affiliation of coauthors per paper (e.g., gender paired team: 3, is a team in which there are 50/50 gender composition in which the authors are from 3 different countries: Colombia, Argentina, and China).

Diversity in teams (in science) can come as separation (members with different life realities), variety (people with different scientific expertise), and disparity (individuals with more/less seniority) (here for more details).

How to instrumentalize this team-diversity factors?

Here, I want to study the case of my country, Colombia, and its scientific outputs and science teams through these lenses.

1. Separation

A variable we can consider studying diversity aspects in science and science teams concerning separation could be the Number of Unique Countries of Institutional Affiliation of Coauthors per Paper. For instance, different countries of institutional affiliation among the coauthors of a scientific paper could display similar or different cultural perspectives or values.

2. Variety

The different areas of expertise among the coauthors could represent a form of variety. For example, there is more variety in a team of three experts from three different scientific domains such as physical, social, and biological sciences, than a team of the same number of authors but just from the domain of health sciences. Thus, a variable that could reflect this property could be Number of Unique Research Field ‘Expertise’ of Coauthors per Research Document.

3. Disparity

Here we consider two well acknowledged problems concerning inequality for women in science: team composition and leadership (regularly, mostly men in both scenarios):

• Team Composition: there is an underrepresentation of women in science teams (there are more men than women in teams). Which might affect the pool of further candidates for tenure-track positions and future research leadership positions, our next variable. Therefore, I propose a Team Composition variable. There are 3 types i) ‘female driven’ is a team where over a half of members are women, ii) ‘male driven’ where over half of members are men, and iii) if they are 50/50 it is ‘gender paired’
• Team Leadership: disparity can also refer to the distribution of power and resources in a team. In this case, the variable of Team Leadership could reflect this power allocation-role based on the author's position. Thus, there are 3 types i) ‘female lead’ where first-last authors are women, ii) ‘male lead’ where first-last authors are men, and iii) ‘gender paired’ where first-last authors are man-women or woman-man.

Visualizing a country’s team diversity

For this first draft of the diversity aspects on separation, variety, and disparity, I used the snapshot April-2024 of Colombian publications indexed in OpenAlex. Also, I used the gender R package to infer authors' gender based on first name.

A few (descriptive) interesting findings:

• Most of the teams are local (N= 12,902) with a substantial composition (N= 6,565) of teams with 2–3 different country affiliation, and where there are just a few truly exceptional transnational teams with over 4 different countries (N= 248).
• It is quite interesting to see that within the latter group of teams over 4 different countries, the majority (~56%) can hold expertise between 2 to 3 different research field expertises.
• As we might suspect, the most common team (N= 1,972 documents) is composed of all researchers affiliated with Colombian institutions with 2–3 distinct research expertise driven and lead by men.
• Which, surprisingly, is more frequent (N= 1,889) than teams only from Colombia of a single expertise driven and lead by men.
• Those uncommon teams (N=4), highly interdisciplinary (≥4 fields) and resolutely international (≥4 countries) are also driven and lead by men.
• Female lead and driven teams (N=570), are local and uni-disciplinary.
• Out of all teams, ~65% were men driven, followed by women (~19%), gender paired driven (~17%). However, of all of those men driven ~25% were gender paired lead, whereas of the women driven ~42% were gender paired lead. In sum, it is more likely for a team to have a gender shared leadership if there are more women .
• Out of all teams, ~51% were men lead, followed by gender paired (~37%), and women lead (~12%). Interestingly, among those men lead teams, ~98% were men driven, whereas those gender paired lead where ~39% gender paired driven and ~21% women driven.

The above and further discussions that follow could illustrate valuable insights into the diversity of teams in scientific endeavors in middle and low-income countries.