Gender(in)g Visualization
Gender diversity in the academic visualization community.
In a recent paper, Tovanich et al. conducted an exploratory analysis on inclusion and diversity in the visualization community by researching visualization done over the years at the IEEE Visualization (Vis) conference. Through this study, the authors found that since 1990, the number of authors at IEE Vis has increased steadily, but the number of female authors grew slower as compared to the male authors. While we will not delve into specifics on the causation of this disparity in this article, this article will help understand where we stand today in terms of diversity at Vis.
This study focuses on data from IEEE Vis conferences over the years, and uses data visualization as a tool for exploratory analysis. Good visualizations need clean data, and to visualize gender representation among authors at Vis, the data has to be as clean as possible! Here’s how the authors of this study cleaned up their dataset:
- Automatic assignment — The authors used some cool libraries in the Python programming language to assign genders based on a pre-existing list of 163,283 unique name + gender pairs.
- Manual Data Cleaning — As hard as this sounds, the authors of the study cleaned up data for authors having more than two publications to their name, either through their own knowledge or based on external web searches.
- Some external help — A double check and an extra set of eyes is always a good thing, so the authors of this study engaged additional folks to to carry out a review on 1514 authors whose automatically-inferred gender had been either manually changed or had remained unknown or unisex.
Here’s a visual representation of the final dataset:
In addition to authors at Vis, the study also looked at various other avenues for gender representation such as the organizing and program committees, community members and awardees. The figures below show some visuals that represent this information quite nicely:
The above visuals provide insights on the number of authors according to gender over the years. While there is an upward trend in the number of authors that identify as female and other under-represented genders, there clearly is a scope for improvement. The study doesn’t dive into the causalities involved in the slower increase for underrepresented genders, but it does help provoke thoughts on how this can be changed.
To further strengthen their analysis and provide additional insights, the authors also included career age (meaning average years of experience), so as to understand the evolution of how soon in their careers do male/female authors publish their work at Vis.
From the visuals, we can deduce that career age on average increased from 3 to 8 for new female authors and from 4 to 9 for new male authors from 1990 to 2020. Barring new data, we clearly see some scope of improvement for new authors and their evolution to publish, as they say, Fail Fast is the new way to go!
Next, the study looked at females being in pivotal roles in publications, meaning whether they were the first, middle or last author in a vis publication.
It is heartening to see that the percentage of females in the first author position has increased over the years from 1990 to 2020! It also is good to see that the probability of more females being the first author as they evolve in their careers is higher!
Next, the study looks at the number of organization committee (OC) and Program committee (PC) members across the years. My personal observation in this visual is that while there is a steady increase in the number of female committee members, we need more data and participation from other underrepresented genders to be able to gauge this information correctly.
Last but not least, this study looks at the award recipients at Vis based on gender and types of awards. While there is a stark difference between the number of female awardees vs male awardees, there is also hope that this will change in the near future, after all, who run the world? Girls!
To sum it up, this study helped portray factual information on gender disparity between the academic visualization communities as of today and paves the way for us to be mindful of these differences and provoke our creativities to help bridge these gaps. Lastly, I hope that future studies will diversify their research to include more data on other underrepresented genders and we will see new and improved results!
References
- Natkamon Tovanich, Pierre Dragicevic, Petra Isenberg. Gender in 30 Years of IEEE Visualization. IEEE Transactions on Visualization & Computer Graphics. 10.1109/TVCG.2021.3114787
