Muddy America: Color Balancing the Election Map
October 18, 2019 12:33 PM   Subscribe

The Trouble with the County Winner Map, and why this Muddy Map is better for determining vote populations and vote margins in the US election.
posted by agentofselection (23 comments total) 26 users marked this as a favorite
Interesting. I like seeing more informative graphics for this sort of info, and this is definitely more informative (if a bit drabber)
posted by rmd1023 at 1:29 PM on October 18, 2019

I've seen the one with the proportionate population representation of blue vs. red. I liked it because it looks like a blue sea creature being deformed and strangled by a discarded red plastic six-pack ring.
posted by Countess Elena at 1:34 PM on October 18, 2019 [7 favorites]

I still like the XKCD 2016 election map for its clarity and simplicity. Like the muddy map, it shows both population density and the mix of votes within each state, though without aiming for the same level of county-by-county precision. But I think it's a reasonable tradeoff of resolution for legibility.
posted by mbrubeck at 1:50 PM on October 18, 2019 [13 favorites]

I like this because it shows that the polarization of the country isn't as extreme as certain actors (CW: 45) make it out to be. The extreme one-sided districts are almost exclusively right-wing and tiny; i's hard to draw the eye away from the large spatial expanse covered by those very few people.

How-not-to-lie-with-maps is a big part of undergrad geography. It's very difficult to get right.
posted by klanawa at 1:53 PM on October 18, 2019 [6 favorites]

another version helping show the difference
posted by kliuless at 1:56 PM on October 18, 2019 [7 favorites]

The map kliuless linked is the best response I've seen to this stupid county map. Land doesn't vote; people vote.

Except not really. Actually electoral college members vote for President. And for some dumb reason in almost every state it's a winner take all vote. So a map of state by state colored red/blue and states sized to number of electoral college votes would be the most direct visualization of a presidential election. Like here or here.
posted by Nelson at 2:07 PM on October 18, 2019 [2 favorites]

I like this map (and the xkcd map) because I think physically twisting or shrinking states to represent population makes things less clear, and also provokes reflexively negative responses in many people who live in states that gets smaller when adjusted for population. If the goal is to inform someone who hasn't already thought much about population density & vote totals, I think muddy > distorted.
posted by deludingmyself at 2:46 PM on October 18, 2019 [1 favorite]

So I guess, if I were a political operative, I'd want to target the dark grey areas? Since they have the greatest number of neutral voters who could potentially be swayed one way or the other?
posted by panama joe at 3:28 PM on October 18, 2019 [1 favorite]

"So I guess, if I were a political operative, I'd want to target the dark grey areas?"

For the 2016 election, and likely 2020 election, those dark grey areas are colloquially called "suburbs".
posted by Ivan Fyodorovich at 3:54 PM on October 18, 2019 [2 favorites]

So I guess, if I were a political operative, I'd want to target the dark grey areas? Since they have the greatest number of neutral voters who could potentially be swayed one way or the other?

My reading of it was that a dark gray county signifies both 1) a relatively large population and 2) a number of Dem voters that is pretty close to the number of Repub voters. I think political operatives would still want to target the dark gray counties, but I don't think the neutrality of the voters is accounted for on the Muddy maps.
posted by 23skidoo at 3:55 PM on October 18, 2019 [2 favorites]

Also, swinging a county doesn't matter much in presidential elections, since every state except Maine and Nebraska bases their electoral votes on popular vote at the state level, am I wrong?
posted by RobotHero at 5:20 PM on October 18, 2019 [2 favorites]

And that's another reason the initial choice to show a majority-per-county map for a presidential election is misleading. It manages to neither show the number of votes, nor the actual results of the election. Presumably for the sake of getting some red into "blue" states.
posted by RobotHero at 5:35 PM on October 18, 2019

This is a nice visualization. I spend a lot of time thinking about how to present multidimensional data, and this is quite well done.

One possible tweak would be to encode vote count into saturation/lightness logarithmically instead of linearly. That would help address the fact that a small number of counties swamp out the range without introducing an arbitrary cutoff, and might help show county-level differences in low-population areas better.
posted by biogeo at 5:38 PM on October 18, 2019 [2 favorites]

Is there a map that shows electoral college vote per capita? I know this is unequal but don’t understand it fully.
posted by q*ben at 6:30 PM on October 18, 2019 [1 favorite]

> Is there a map that shows electoral college vote per capita?

Here you go: Map (FairVote)

Wyoming, Alaska, Vermont, Rhode Island, and North Dakota are the Big Winners.
posted by flug at 6:36 PM on October 18, 2019 [1 favorite]

It seems like the color should be scaled by population density, not absolute population. The red chunk around Las Vegas (I think?) looks like more red votes than the blue chunk around Los Angeles, but that's not true.
posted by scose at 7:26 PM on October 18, 2019 [2 favorites]

I decided to extend my analysis of the electoral map and Trump's counties from the other side: look at the most rural counties and see how much land and votes Trump and Hillary won.

I made my cut-off at counties with a population of less than 1,500. There are sixty-one of them and that seemed sufficient to say something. Since Alaska doesn't report its votes by boroughs (their equivalent to counties), I relied on a paper that studied boroughs versus voting reports in Alaska from 1960 to 2016.

Trump won 54 of these counties and Clinton 7. Among Clinton's victories were Native-American dominated counties, and the smallest and least populated county in the U.S.: the former leper colony that is Kalawao County, Hawaii. Five square miles and 88 people. Hillary got 14 votes and Trump 1.

Of course the real story about the rural counties is that Trump won 54 out of 61. His total vote in those counties is 23,664 compared to Hillary's 3,910, outdistancing Hillary approximately 7 to 1. For those 23 thousand-some votes, Trump got bragging rights on 65,572 square miles or about three square miles per vote.
posted by dances_with_sneetches at 7:48 AM on October 19, 2019 [2 favorites]

The creator of the map linked by kliuless has posted an updated version showing the proportion of the vote for each county instead of winner-take-all. I think this is almost ideal, though I'd prefer to see a tint rather than the sort of horizontal pie graph he's using.

It should be possible for someone who knew what they were doing to edit it. (I am not that person.)
posted by designbot at 4:26 PM on October 19, 2019 [1 favorite]

If preserving county boundaries is critical, then the Muddy approach makes sense.

The price is that it discards a lot of population information. Every county above 59828 votes gets the same full saturation -- whether that's 60K or two million is invisible. Maybe I can find the table of county votes and figure what % of all votes are invisible this way.

I don't recognize most of the counties, so for me the xkcd approach communicates more.
posted by away for regrooving at 11:20 AM on October 20, 2019

The Electoral Vote decided the presidential election, and it's by states, not counties. The county maps are an interesting at where people actually live and vote, but they don't show how the election was decided.
posted by kirkaracha at 6:59 PM on October 20, 2019

One possible tweak would be to encode vote count into saturation/lightness logarithmically instead of linearly

They actually discuss this in the article. When they changed to logarithmic instead of linear w/a cap, it was harder to differentiate the saturation between counties with less votes. Those 2-4 counties that skew everything in the linear map also skew things too much in the logarithmic one too.
posted by LizBoBiz at 4:54 AM on October 21, 2019 [1 favorite]

scose has a good point regarding population density vs. total population. If you use total population then a large county can feel like it has more voters than a smaller county with the same population. Because lightness (or inverse in this case) does naturally map to population density. If you put the same lightness across twice the area, you intuitively think it will be twice the population.

Like, imagine you fill a paper with dots. That's a direct representation of population density, right? Then do the same thing but with more dots but smaller. Keep repeating with smaller dots until you can't see individual dots, you'll get a grey paper. So it's natural for lightness to show population density, compared to hue or anything else.

And a linear scale also makes sense with that. If you put twice as many dots, it will be twice as much "ink."
posted by RobotHero at 8:17 AM on October 21, 2019

In Canada we just had an election, and here's one approach for displaying results without overvaluing large low-density land. This doesn't account for the margins of victory at all, but that could be complicated by having to account for more than two parties:
posted by RobotHero at 9:30 AM on October 22, 2019

« Older Helping the climate survivors multiply and thrive...   |   No Claws for Concern Newer »

This thread has been archived and is closed to new comments