It does explain maybe 20% of it. It's because it's tracking cumulative deaths.
Say there's 100 Trump counties and 100 Biden counties.
80 Trump counties are in the South/NW. 80 Biden counties are in the West/NE.
West/NE rack 60% of their deaths in the Winter. South/NW rack 40% of their deaths in the winter.
If the timeline is cut before winter in this example using almost completely arbitrary numbers then this is what happens.
Of the cumulative deaths for any given year the majority of Trump counties have 60% of the years deaths before winter and Biden counties have 40%. When calculated with the random 80/20 regional split this gives us...
(0,8x60) + (0,2x40) = 48 + 8 = 56% of the yearly cumulative deaths in Trump counties before winter
(0,2x60) + (0,8x40) = 12 + 32 = 44% of the yearly cumulative deaths in Biden counties before winter
Regional weather effect isn't a huge effect, but it does exist and if you're looking to make a point then choosing this timeline from the start of the year does "favor" Biden counties.






Reply With Quote