The Problem of Analysing the US 2016 Presidential Election

Posted on the 30 March 2019 by Doggone
There's a big problem with trying to analyze the US 2016 Presidential Election.
TRUMP WASN'T POPULARLY ELECTED
He was elected by the Electoral College, which isn't tied to the popular vote.
As I like to point out, that Hillary Clinton had a significant amount of the popular vote kind of trashes most of the "she was unpopular" analysis. Since she WAS popular: just not where and how it counted.
So, talking about what the Clinton campaign and the Democrats did to lose the election seems wrong. Unless you are doing it in a strategic manner.
The Clinton campaign thought that no one would vote for trump. That led to the assumption that most of the brown states in this map would vote for Clinton. That's why there were all the predictions of a "landslide" for Clinton.
She could have pulled off a win if she had taken Florida and any one of AZ, MI, NC, PA, or WI, but she didn't. But that isn't really about the popular vote as much as it is about the workings of the Electoral College and failing to realize that a "safe" state may not be that safe (e.g., MI, PA, and WI).
I'm going to add that I learned that Andrew Therriault, the former DNC director of data science, reacted, “Irony of her bashing DNC data: our models never had MI/WI/PA looking even close to safe. Her team thought they knew better.””
The real upshot of all this is that what counted were the Electoral College votes and getting enough to get all a states votes: since Clinton probably would have won if the Electoral College votes somehow reflected the popular vote.
But the real bottom line assumption was that there was no way she would lose to Donald Trump.
Boy, we're they WRONG.
See also: