Read on Twitter
A crucial point that data-driven journalists need to heed before their headlines drive any narrative too far. Counties are terrible units of analysis for interpretations of the election. Two quick points to add to this about the ecological fallacy: https://twitter.com/matt_blackwell/status/1328404656914108416
1) We KNOW the ecological fallacy is working here (its not ALWAYS the case that aggregate!=individual), so really don't make claims based on this unit. Wait for or use individual-level data! https://www.washingtonpost.com/politics/2019/11/13/white-trump-voters-are-richer-than-they-appear/
Even if the ecological fallacy was less dire, they're too different as units for a good comparison! ~3200 counties in the country. 1/2 pop in ~70 of them. Bottom 600 counties combined have ~1%.
It's even more extreme when we look by economic activity:
Punchline is almost all the "rich" people (1% and up) in the country live in like 70 counties. So does 1/2 of all US! Including people at all places in the income distribution. Big counties just look richer on average because the smaller one don't have any rich people.
To punctuate: the smallest 1900 counties sum to about 10% of total pop. The biggest 130 are 50%. So a county-level analysis can easily tell us almost nothing about "how Americans voted" because it weights those empty places so much and oversimplifies big ones to 1 number
