I rather doubt Adams gave Clinton a 0% chance. There was always some self doubt in his prediction. Other posters have pointed to Adams briefly suggesting Clinton was favorite. This does not suggest a 0% chance of a Clinton win. And no, I am not interested in any flippant quote from Adams taken out of context. I watched and read enough of his output during the election to know he did give Clinton a chance. I am witnessing here multple posters having a go at Adams for contradictory reasons. You can’t all be correct.
With the object of accuracy in mind it’s worth pointing out Silver gave Clinton not a 70% chance of winning but a 71.4% chance. For a while on election night this increased to a smidgen under 75% chance of a Clinton victory. If you are proposing accuracy lets then put out accurate figures, no mass rounding down to make your argument look better.
At the same time that Silver’s model was giving a 75% chance of Clinton victory, he was also pointing out that it wasn’t taking key data into account, and that the New York Times model which did take that data into account showed her chances as much lower.
And it’s tough to say just what percentage Adams was giving Clinton, because he never actually said any percentages. That’s another failing of him as a prognosticator: He didn’t give numbers.
If if and buts were candy and nuts, we’d all have a merry Christmas.
It was because of the 9/11 memorial service event that Adams said Trump was “running unopposed” - he made that prediction based on what really did happen. And he was wrong. The race was close, as, in fact, it was always going to be.
I apologize but I have to say this; if you think Silver did not take into account risk, you literally do not understand what he was doing or what the numbers mean.
If he did not take risk into account, his prediction would have been that Clinton was 100% likely to win. If he gave Trump any sort of a chance beyond Lloyd Christmas odds, then, by definition, he was accounting for risk. That’s what “risk” is.
Sorry, but I don’t think you are quite correct. The potential of Trump running unopposed was discussed by Adams long before the 9/11 memorial service.
It was always going to be close? No, it wasn’t. Imagine Comey had recommended indicting Clinton the weekend before the election. How close do you think the election would have been then? Im imagining at least 30 to 40 extra EC votes going to Trump in that scenaro. It’s then a landslide.
edit: in my earlier post I should have said Silver did not take into account the element of risk surrounding a Clinton health event or scandal(or he didn’t weight it very well). The main risk Silver seems to have taken into account was the risk that the polls were wrong.
Perhaps it was. But once Clinton collapsed, Adams said “Trump is running unopposed.” He stated it as fact. How often does Nate Silver just come straight out and say something like that? Never, because he understands risk, probability, and chance.
The polls always held that the race was close. Clinton never held a double digit lead over Trump in the poll aggregation numbers, and the spread of votes, as Silver’s prediction noted (and as happened) tilted the election towards the Republican candidate by a couple of points, since the Democratic candidate needs a point or two extra to get an EV victory.
Well, yeah.
Of course there is risk Clinton could collapse, or be indicted, or have a heart attack. But there is surely equal risk something could happen to Trump. He’s a fat, unhealthy man; HE could collapse or have a heart attack, or something like the Access Hollywood tape could have popped up because God knows Trump has a lot of skeletons in his closet. Unless you have a reason to believe one candidate is more likely than the other to be hit by a meteoric event, why would that alter the odds? And we had no such reason, so the odds even out.
Nate Silver is first and foremost an egghead. I wouldn’t expect him to make the odd outlandish or flippant claim that Adams’ makes. Adams is someone I’d expect to do that. That’s not meant as a criticism of Adams, it’s just how some prefer their political punditry; some analysis, some punditry and some thinking outside the box, some statements said off the cuff that you might just take back at a later date.
It was the opinion of Scott Adams that the risk of a catastrophic event for each candidate was not even. It’s one of the reasons he said Trump would win, and likely win in a landslide. He placed the catastrophic risks to Clintron as far higher. Adams said Clinton didn’t appear healthy to him. He reasoned this by her physical appearance. Adams said her appearance and health seemed to change from day to day. This to him was a sign of less than optimum health for someone in their sixties. He also said Clinton was under greater risk of a scandal - the FBI investigation and Assange was circling Clinton on Twitter with comments about an upcoming leak. All these things combined led Adams to say the risks were far greater towards Clinton. Now, you may not agree with his assessment, but that is how Adams explained it. As it turns out Trump had his own risk of a catastrophic event. I would say both candidates almost received near knock-out blows for their respective campaigns, Clinton twice, Trump once. Does this all mean Adams was correct in his breakdown of risk? Im damned if I know, but Adams has got me interested in his opinions - for now.
The risk of scandal for Trump was 100%: Multiple scandals were already known and completely confirmed. Clinton’s chances couldn’t possibly be higher than that.
And on the health front, we don’t know just how bad Trump’s health is, because he falsified his doctor’s report (itself already a bad sign), but we do know that he’s a man, older than Clinton, who shuns exercise and eats unhealthily. There’s no rational basis on which to decide that Trump is healthier than Clinton.
Meanwhile, Silver does actually consider the possibility of unforeseen catastrophic events, and incorporates them into his model.
He doesn’t (because he can’t) look at specific unforeseen events, but he can and does consider such events in the abstract. For purposes of the model, you don’t need to know the specifics of the event, just things like the probability of an event that causes a swing of 10 points (or 5 points, or 2 points, or whatever) one way or the other. One consequence of this is that his method will never give a probability higher than about 95% more than one month out, because the possibility of an unforeseen Big Swing event in the last month of the race is around 5%. But the probability for the favorite will generally creep upwards in the last month, as time passes and that Big Swing event fails to materialize.