What 2016 should have taught the US...
From that link:
The site Five Thirty Eight projected that Hillary Clinton had a 71.4% chance of an Electoral College victory... The New York Times projected an 85% chance of a Clinton presidency.
That means they were predicting 28.6% and 15% chances of Trump winning.
Here's a graph from "538" showing that they had the win/loss odds going up & down at a fairly constant interval over the period leading up to election day, with election day being at the lowest point in the cycle, predicted at 64½ Hillary and 35½ Trump right before the election.
Here's another showing the distribution of their predictions under different models, showing how close the dividing line between a Hillary win and a Trump win was to the middle of the range ahead of time.
Here's an article they published in September that year titled "Democrats Should Panic If The Polls Still Look Like This In A Week", which includes a map showing each of the states that failed us in blue, but pale blue, close to a white toss-up, and it has their borders in a thicker black line than most other borders, which the map key shows was their label for states they identified ahead of time as "tipping points". The information was there all along. Those who paid attention to it were warning us all along. Those who said everything's fine were simply not using it, not basing anything they said on the facts.
Ironically, I bumped into those graphs (my first two links of the three) while searching for something else, and they happen to have ironically been embedded in
this ironical election-day Huffington Post article ironically titled "What's Wrong With 538" which ironically actually shows both that 538 had gotten it right while HP had gotten it wrong, and also how.
Given the fact that all of the article's data showed 538 being right and the article's author(s) being wrong, what was the basis for their conclusion going exactly the opposite way from the data, on election day but before the results were in (and treating Hillary's obvious win as an inevitable foregone conclusion)? If you read the article, you can see their "reasoning" all over the place in plain English, so there's no need to try to infer it. Each time they show a data point they don't like, they just answer it with "That leads to this conclusion. This conclusion obviously can't be right, so pointing to it proves them wrong". You couldn't ask for a more direct, obvious, textbook example of starting at the conclusion. They're literally telling us themselves "We start at our conclusion!".
But at least that one showed the data. The link in your post is mostly notable for its lack of it and biased filtering of it. The only thing it had other than what I've already quoted was a few survey percentages in certain states, showing not how big Hillary's supposed advantage was on election day or as an average over the previous x months but simply the biggest they'd ever peaked at at any time over the previous several months. The entire rest of the article was just narrative. So what's the source for the conclusion? That was it right there: narrative alone, with nothing else there for it to even possibly have come from. Again, that's a demonstration of the process: ignore data, just stick to the approved narrative as proof of itself. It's the only way to get back to the conclusion you started at.
I was going to compare the survey numbers in the states that mattered with the election results, but I'm out of time. For reference,
here's the election outcome data. Look how wildly hugely different it is from the predictions! Sometimes up to a whole couple of points!
We think alike. I ask at the beginning if it is a push poll, they say no, and as soon as it's obvious it is I explain to the poll taker (because they are usually just hired callers) what a push poll is and hang up.
