The day before the 2016 US Presidential Election, most pollsters and statistical models had pegged Hillary Clinton’s chances of winning at greater than 90%.
However, as we noted yesterday, the consensus view is not to be trusted in a post-Brexit world.
Here’s what went wrong:
We looked at the predictions made by 12 major newspapers and pollsters the day before the election, to see where they went wrong.
[drizzle]For Wisconsin, Michigan, Pennsylvania, and New Hampshire – not a single source gave an edge to Republicans.
| Correct | Tossup | Wrong | Fail% | |
| Pennsylvania | 0 | 0 | 12 | 100% |
| Wisconsin | 0 | 0 | 12 | 100% |
| Michigan* | 0 | 0 | 12 | 100% |
| New Hampshire* | 0 | 1 | 11 | 92% |
| Florida | 1 | 3 | 8 | 67% |
| North Carolina | 2 | 2 | 8 | 67% |
Note: at time of publishing, Michigan and New Hampshire are very close and leaning Republican. The point still stands. Every pollster said these states were “likely” or “leaning” Democrat.
For Florida and North Carolina, the pollsters were slightly less reckless. The Associated Press correctly had the Sunshine State as “leaning red”, while the Huffington Post saw North Carolina ultimately voting Trump.
After this and the Brexit polling disaster, the media is sure to be much more cautious with their models going into the next big political event.
By Jeff Desjardins, Visual Capitalist
[/drizzle]
