It is highly likely that people who choose to download a Covid symptom tracking app are more concerned about Covid than the average person. In that case they are likely to take more precautions than average to avoid catching or transmitting it.
What might that mean about the representativeness of symptom onset trajectories from such a group? Will they track the whole population trajectory, or do something different?
Sharpening this question as follows:
What do the daily infection trajectories look like for the 10% (say) of the population with the lowest individual transmission probabilities, and how do they compare to the infection trajectories for the whole population?
Simulations visualise and reveals the hazards of this opportunistic sampling that may be explained by self-selective bias or biased sampling.
SEIR
The final version: WG09P2.R draft file: code.R