Sunday, November 08, 2020

The demographics of the US election

There was a fascinating analysis of US voting demographics in the Globe and Mail this weekend (based on Associated Press' data).

It reveals the extent to which Joe Biden can thank women and black people for his slim victory, and the extent to which white Americans, particularly white male Americans, still find themselves able to support Trump and Trumpism, despite the unmitigated disaster of the last four years.

Overall, 55% of women voted for Biden (compared to 44% for Trump), while only 46% of men voted for Biden (52% for Trump). When we start to take ethnicity into account, the discrepancy becomes much more stark: white people voted overwhelmingly for Trump (59% of white men, 52% of white women), while black people voted even more overwhmingly for Biden (87% of black men and a huge 93% of black women), with Latinxs somewhere in between (59% of Latinos and 66% of Latinas voted for Biden). 

The other analysis that sticks out like a sore thumb concerns where people live. Urban voters overwhelmingly voted for Biden rather than Trump (65%-33%), while rural voters overwhelmingly voted for Trump rather than Biden (also 65%-33%). Suburban and small town voters fall neatly and logically between those two extremes. This is the main reason the American electoral map shows such a marked pattern, one that is largely repeated election after election, with the east coast and west coasts and the north solid blue, and the more sparsely-populated centre and south of the country mainly red (although this has started to change to some extent, as ethnic demographics change and states like Arizona and Georgia start to turn blue).

Conpared to these demographics, considerations of income and education levels pale into insignificance, although generally speaking the more educated the more likely people are to vote for Biden, and lower income and higher income earners are more likely to lean Democrat, and middle income earners more likely to lean Relublican.

No comments: