💡 Takeaway: Family ties, especially shared households, are key drivers of infection.
But schools can still amplify small increases in risk.
✅ Rich registry data helps us pinpoint how SARS-CoV-2 really spreads.

Full paper + code:
🔗 doi.org/10.1038/s415...
🧪 github.com/jgarciab/cov...

But take the results with a grain of salt: schools were closed for ~1/3 of the study period, and we only studied PCR test results (i.e., no asymptomatic or mild cases).

We also found that most variation in transmission risk stemmed from individual and school-level factors, not geographic location.

What about classmates?
We matched students by school, program, and prior shared classrooms. Here's what we found:
Same program track (G4): 1.65%
Same school, diff. program (G3): 1.11%
Same primary, diff. secondary (G2): 0.66%
No school overlap (G1): 0.52%

2/
Family ties seem to matter far more than school ties

Probability that two siblings test positive within 14 days: ~23%
For students attending the same school: just 1.6%

💡 Even living in separate homes, family members had higher probability of testing positive at the same time than classmates.

In summary, while unsigned data may be enough to assess ideological polarization, they fall short of identifying extreme users engaging in antagonistic behavior, only possible with signed data.

Thanks to S. Babul (first shared co-author), Ö. Togay, @abovet.bsky.social , and @javiergbe.bsky.social