Very interesting! I'm a bit confused about the data/graph that shows their main conclusion, anyone know why: They use a 7-day rolling average and general conse

Very interesting! I'm a bit confused about the data/graph that shows their main conclusion, anyone know why: They use a 7-day rolling average and general consensus is that covid has a 3-14 day incubation period. Combining these two would indicate that the mask's effect should be gradually becoming stronger starting ~4days after introducing and reaching peak effect after 21 days in the graph. The graphs show a strong effect after 4 days and remain constant after that. Does that undermine the conclusion or am I seeing this wrong? They use a linear regression while infection is an exponential process in my understanding, doesn't this undermine the regression?