Before we all get caught up in the hysteria, it's good to be reminded just how thin the margin was.
You Could Fit All the Voters Who Cost Clinton the Election in a Mid-size Football Stadium
While nearly 138 million Americans voted in the presidential election, the stunning electoral victory of Donald Trump came down to upsets in just a handful of states that Hillary Clinton was expected to win. It has been cold comfort for Democrats that Clinton won the popular vote—at the last count, she was up by about 2.5 million votes, and climbing, as ballots continue to be counted. Even more distressing is the tiny margin by which Clinton lost Wisconsin, Michigan, and Pennsylvania—three states that were supposed to be her firewall in the Rust Belt, but that ultimately tipped the electoral college map decisively in Trump’s favor.
Trump’s margin of victory in those three states? Just 79,316 votes.
This latest number comes from Decision Desk’s final tally of Pennsylvania’s votes, where Trump won 2,961,875 votes to Clinton’s 2,915,440, a difference of 46,435 votes. Add that to the official results out of Wisconsin, where Clinton lost by 22,177 votes, and Michigan, which she lost by 10,704 votes, and there you have it: 0.057 percent of total voters cost Clinton the presidency.
It is not entirely unusual for the electoral college to be lost by such a slim margin. In 2000, Al Gore lost Florida (and therefore the election) by 1,754 votes, triggering a painfully drawn out recount drama that only ended with a Supreme Court ruling. And in 2004, John Kerry lost to George W. Bush by losing Ohio by a little over 118,000 votes. But it is worth considering just how few voters ultimately set the country on its current, arguably terrifying course. The 79,316 people who voted for Trump in Wisconsin, Michigan, and Pennsylvania—all states that Democrats carried since 1992—is less than the entire student body of Penn State (97,494 students), or only slightly more than the number of people who attended Desert Trip, the Baby Boomer-friendly music festival colloquially known as “Oldchella.” If you put all these voters in the Rose Bowl, there would be slightly over 13,000 seats left over. There are more people living in Nampa, Idaho, a city you have never heard of.
Mark:
ReplyDeleteThe above quote is slightly misleading, in that if there were a shift of 80,000 votes in those three states, you'd expect to see another million or so votes shifting all across the country. It's not like those votes would shift on their own.
Well, yes and no…
DeleteObviously it would be silly to treat the results in one state or region as independent from another and we generally expect the correlations to be positive. That said, we are talking about small numbers here and we can allow for the possibility of something like a slightly lower turn-out in Southern California offsetting small but sufficient gains in the Rust Belt. (Lots of possibilities – earthquakes, Santa Anas, fires, or the thing that really freaks out the locals, light rain.) Not at all hard to imagine scenarios where Hillary loses some margin on her popular vote victory but wins the EC.
We could have an interesting conversation about reasonable distributional assumptions. The trouble is that is not the discussion that people with the big platforms are having.
For example, we’ve seen a major surge in the argument that Trump is unstoppable in large part because the Republicans have a huge edge in the electoral college largely due to their advantage in Midwestern swing states. This is where, I believe, these numbers are relevant to the debate, because it is difficult to come right out and argue that 80K votes spread out over multiple highly populous states constitutes a firewall.
It is, of course, entirely possible that the GOP will hold every state that they won in 2016, but much of the conventional wisdom that’s pushing this line among the data journalism crowd ignores some inconvenient numbers, and the public discourse would be better if everyone starts addressing the relevant statistics.
Yes, I discuss some of that here: https://statmodeling.stat.columbia.edu/2016/12/18/bernie-sanders-lost-presidential-election/
DeleteThe first sentence of the post is called, "Nobody knows what would’ve happened had Bernie Sanders been the Democratic nominee in 2016."
I guess with this kind of attitude I'll never be a successful pundit.