Social Data Science·University of Copenhagen
Understanding society and science through population-scale data and computation
Andreas Bjerre-Nielsen
SODAS·
In much of my research I work with populations rather than samples. Using four decades of Danish administrative registers, I study how education systems, neighborhoods and social networks shape life outcomes, and how algorithms and AI are beginning to reshape both public decisions and research itself.
I am based at SODAS, CAISA and the Department of Economics at the University of Copenhagen. At CAISA I lead the research collaboration with Statistics Denmark.
Featured Work
All papers →- 2025
Removing Phones from Classrooms Improves Academic Performance
In a randomized trial with nearly 17,000 students, we find that collecting phones during class raised grades, with the largest gains among lower-performing students.
- 2025
The Origins of Large-Scale Structure in Family Networks
We find that what holds Denmark's family network together is not that partners resemble each other but that people separate and re-partner: each new couple is a shortcut that pulls distant parts of the population within a few steps.
- 2024
Trading off performance and human oversight in algorithmic policy: evidence from Danish college admissions
We find that machines rank applicants ahead of admission both more accurately and more fairly than the policies in use, beating grade averages and human assessment alike, though at some cost to human oversight.
- 2024
Attendance Boundary Policies and the Limits to Combating School Segregation
We show that redrawing school district boundaries failed to change schools' socioeconomic composition, because households reassigned to lower-status schools moved or opted out instead.
- 2022
Rising temperatures erode human sleep globally
We find that warmer nights already shorten sleep worldwide, mainly by delaying when people fall asleep, and that the loss falls hardest on older people, women and residents of lower-income countries.
- 2021
Task-specific information outperforms surveillance-style big data in predictive analytics
We find that digital trace data predicts outcomes best when it captures behavior specific to what is being predicted, rather than general surveillance of daily life.
Research Themes
AI and technology
What prediction systems actually add, where they fail, and how they change the way we measure people.
Algorithms and allocation
How the rules that assign school and university places shape behavior, access, and who gets what.
Education and human behavior
School choice, admissions, and the effects of technology on how students learn and perform.
Networks and neighborhoods
How networks and neighborhoods shape opportunity, from peer effects to large-scale social structure.
News
All posts →- 15 Feb 2026
Part-time research advisor at Statistics Denmark
I have taken up a part-time position as research advisor at Statistics Denmark, alongside my roles at SODAS and CAISA.
- 07 Nov 2025
CAISA and Statistics Denmark to build national infrastructure for responsible AI
CAISA receives 45 million DKK from the Danish research reserve for a new collaboration with Statistics Denmark, which I will lead.
- 25 Jul 2025
Removing Phones from Classrooms Improves Academic Performance
Our working paper on the academic and behavioral effects of banning smartphones in classrooms is now available as an SSRN working paper.
Contact
Offices
SODAS·CSS 1.2.25
Department of Economics·CSS 26.1.14
Øster Farimagsgade 5, 1353 Copenhagen