This is a key paper in the growing literature on the effects of COVID-19 on academia. It is done purely with webscraped data, and it is a showcase how you can do fine-grained analyses if you put in the work.

But:
- Barriers to productivity remain persistent for women scientists
- Yet, the productivity gender gap did not widen during COVID

📖 What did we find?
- During COVID-19, scientists published about 33% fewer papers
- Dutch science was hit hard: “lost” around 20,000 publications
- Scientists with childcare responsibilities struggled to maintain productivity
- Researchers with a migration background were also hit harder

🔭 We analyzed publication data from more than 8,000 scientists across all disciplines at all Dutch universities, covering the period from 1990 to 2022.

🧐 We asked whether women in Dutch science show a stronger productivity decline than men during COVID-19

I built the website with Quarto, R, GitHub, and host it through Netlify. Curious how it’s made? The source code is available on my GitHub repository: github.com/bhofstra/aca.... I licensed it through CC-BY 4.0, so feel free to re-use what you need.

You can find the paper here: www.sciencedirect.com/science/arti...

You can find the replication website to the paper here: bhofstra.github.io/netsize_dutch/

(PS: fitting that this gets out while I'm currently at the Sunbelt conference of the International Network of Social Network Analysis!)

Additionally, we find a pretty strong presence for gender and educational homogeneity.

These findings have important implications for Dutch social structure as to who is at risk to become socially isolated and as to where social barriers between groups exist.

We find fascinating patterns where average network size across a representative set respondents in the Netherlands is about 518 people. But more importantly, it decreases with age, unemployment, lower income, lower house values, lower education, and fewer people in a household.

This is a project that came out of our earlier popular publication at Quest where they asked us similar questions!

In this paper we use computational methods (for insiders, specifically the Network Scale-Up Method) to calculate both network size and its determinants, as well as network homogeneity and its determinants.