On Saturday January 21, 2017, the day after Trump’s inauguration, thousands of people protested around the US (and around the world). Previously I looked at cities with the largest protest crowds, adjusted by city population. Now I’ll compare statewide protest crowd totals to the election vote tallies.
Specifically, I ask the question: What is the relationship between a state’s 1/21/17 protest total and the number of people who voted against Trump?
Again, I’ll use data gathered by Jeremy Pressman and Erica Chenoweth, taken from this site on 1/24/2017:
https://docs.google.com/spreadsheets/d/1xa0iLqYKz8x9Yc_rfhtmSOJQ2EGgeUVjvV4A8LsIaxY/htmlview?sle=true#
I clean up this data a little, take out non-US protests, and group protests by state.
Our 2017 National Election data comes from David Wasserman and the Cook Political Report, compiled here:
https://docs.google.com/spreadsheets/d/133Eb4qQmOxNvtesw2hdVns073R68EZx4SfCnP4IGQf8/htmlview?sle=true#gid=19
I combine the votes for “Clinton” and “Others” into “NonTrump”, organized by state.
Washington, DC, is an outlier because many people came from out of the city. The ratio of protesters to population is 1.976!
Here is a plot of StateCrowd versus Non-Trump Votes, on a log10 scale, which allows you to see which state is where by hovering.
We are interested in fitting a line to the data.
The fitted slope is 0.9375, which means that if State B has ten times as many non-Trump voters as State A, we would expect the crowd to be 10^(0.9375) = 8.66 times larger.
A non-log plot also shows the correlation.
The slope for this fitted line is 0.06997, which we can interpret as saying that on average 7.00% of non-Trump voters attended the 1/21/2017 protest. (Except that the crowd estimates include non-voters, for example children.)
If we remove DC, we can more reasonably compare the 50 states. Here is a heatmap showing each state’s ratio of protest crowd to non-Trump votes.