Let's talk minority voter suppression for a moment, and in "honor" of Cambridge Analytica, let's use some data to back it up. Specifically, let's use the MIT Science Lab's Election Data and County Presidential Election Returns 2000-2016 and the US Census estimates for 2016
Some of you may have seen @Channel4News discussing the Trump campaign's use of Cambridge Analytica to target Facebook and other ads to black voters for "Deterrence." The question you may ask is, "How do I know if his is true or if it even worked?"
There are three strategies to winning an election in the US: get everyone to *want* to vote for you, get people who don't like you not to vote, and get people who don't like you to waste their time voting for a 3rd party candidate. Those last two would fall into "Deterrence"
Take the last one first: third-party candidates. Most of you probably remember Jill Stein's worthless run for the presidency. You may also remember she made regular appearances on Russian state-run RT News. Hopefully you've forgotten her 90's folk-band recordings. *shudders*
But 3rd-party candidates in the US get nowhere, so what's the point, right?

If we look at the election data and compare how the candidates fared in counties where Trump won and where he lost, something interesting emerges:
In counties where Trump won -and lost - the Democrat's average percentage of the vote dipped from the previous year - and the 3rd-parties ('other') went up. A lot. Nearly tripled, in fact from 2% to almost 6% among Clinton-won counties :
But what about the more than 200 counties Trump "flipped" blue-to-red from Obama? Even worse:
OK, so what about the other "Deterrence" effect: get them not to show up? Could social media and negative advertising *really* make that much of a difference?
In a word... yeah.
Bring in the Census data! (that thing Ross is trying to kill right now).
It may be no surprise that the mean per-county percentage of African-Americans in Trump-winning counties was about 1/3 that of Clinton's. What may surprise you is that the black vote in Trump's "flipped" counties is significantly higher than his other counties
In fact, they go as high as 49% black. If you look at those "flipped" counties by how many people actually voted *on average* you could be forgiven for saying no voter suppression occurred - in fact it *looks* like voting went up - just not for the Ds. But averages are funny...
Someone once told me: You can hide a lot of sins in an average. What if we look at Trump's "flipped" counties where over 30% of the population is African-American?
How about 40%?
It should be clear what's going on: Though we can't tell which votes were cast by black voters, the higher percentage, the more the total vote count in those counties fell - as did the votes for Ds - and 3rd-party votes can't explain it.

R votes didn't change. Think about that.
Trump did not flip these counties by getting more republicans to vote for him, he flipped them by getting *fewer people to vote* - especially minorities.

A GOP strategist once wrote about this in chillingly casual terms:
https://www.npr.org/2020/01/05/785672201/deceased-gop-strategists-daughter-makes-files-public-that-republicans-wanted-sea
This is voter suppression. This is why the Postmaster General doesn't want to deliver ballots, and why Wilbur Ross doesn't want a complete census, and why some have been trying everything they can to make voting harder. If everyone votes, they lose. Democracy is inconvenient.
Look, I know you've heard this a million times, but VOTE! If it wasn't important - if it didn't make a difference - they wouldn't spend MILLONS to stop you from doing it!

It is your right. Don't lose it.
PS - When I get a chance, I will publish a full GitHub repo with all the data sources and calculations spelled out so you can check (and maybe improve) my work.
You can follow @jimeharrisjr.
Tip: mention @twtextapp on a Twitter thread with the keyword “unroll” to get a link to it.

Latest Threads Unrolled: