Contacts in context: large-scale setting-specific social mixing matrices from the BBC Pandemic project

Petra Klepac; Adam J Kucharski; Andrew JK Conlan; Stephen Kissler; Maria Tang; Hannah Fry; Julia R Gog

is the dominant eigenvalue of C (its spectral radius). Analogous to the eigenvector representing stage-specific contributions to overall population reproduction in demographic theory [3] , the dominant eigenvector here gives an indication of which age-groups most contribute to transmission in the population, assuming no age-specific differences in susceptibility or infectiousness (Fig 2) . We repeated these calculations stratified by type of contact (physical or conversational) and by context (home, work, school, other), resulting in 8 matrices each for the raw contact matrix shown in Fig S1 and population contact matrix in Fig 3. We further stratify these matrices by contacts made during the week ( Fig S2) and during the weekend ( Fig S3) . As participants filled out the contact survey at the end of the 24 hour period, we define the weekend based on the time they activated the app in the following way. To avoid weekend/week overlap, we consider participants who activated the app after 18:00 on Fridays and before noon on Sundays as those reporting their weekend contacts, and those who activated the app after 18:00 on Sundays and before noon on Fridays as those reporting contacts made during the week. This excludes 2,451 users with 24,983 contacts that report their contacts both during the week and the weekend.