3 ms·

Hypothesis: people who are stuck at home have more time and inclination to check their temperatures, even when they don't feel sick.

• maxerickson 7y ago
  That wouldn't generate an unusual number of 'fever' readings, and then a pretty simple model would do some accounting for repeated use of each thermometer (so 'extra' normal readings shouldn't be reducing the expected number of 'fever' readings).
  • rossdavidh 7y ago
    Agreed, but it could easily generate an unusual number of "not a fever" readings, which would make it look as if there were a smaller than normal number of people getting sick with the flu. Or, maybe it doesn't do that. I don't see how you can say for certain which way it would shift, but I find it implausible that it would stay the same.
    • maxerickson 7y ago
      That's what I mean by a simple model. Like a dead stupid one would be to compare to historical data for fevers/week per thermometer. That is barely a model at all and it manages to ignore how many readings are done and accounts for multiple readings during a period of illness and readings for multiple sick members of a household and so on.