The 2016 polling actually wasn’t that far off: The popular vote ended up being similar to what pre-election polls had predicted, but the Hillary campaign stumbled in the Rust Belt swing states.
Generally, how much of a lead does it take for you to feel safe? Suppose Biden were only 5% ahead of Trump (in both swing states and in national popular vote,) do you feel good? What about percentage likelihood of victory - do you feel good at 70% chance of win?
I won’t feel safe until the votes are counted. But, a lead of 5% of more in the battleground states would reduce my anxiety somewhat.
538 gave Hilary a 70% chance at winning and I was nervous as hell before results started coming in. It got worse from there.
If I wasn’t worried about other shenanigans it would probably take a 538 >99% for me to feel safe. Like, Trump is actively getting impeached with GOP support and throwing objects at reporters, dropping N-bombs in public levels of insanity.
Today Trump threatened to withhold federal aid from two states if they go ahead with mail-in ballots (actually it was for even less than that). I fully expect voter suppression at the least and throwing out ballots/actively changing totals at the worst. I won’t feel safe until Biden is sworn in.
Your poll is missing options, namely “I won’t feel secure no matter what.”
I won’t really feel safe until the Senate announces the electoral vote total. I actually felt safe in 2016 until Comey put his fist on the scale, then it was pure terror.
I had considered that, but didn’t want to include it because I felt it might become something of a cop-out option. I wanted posters to actually have to put a number to their feelings.
I was about to say. Obviously 70% wasn’t high enough to be safe. Maybe 90%.
I would feel mostly safe if, in late October, 538’s model has the Democrat as at least 90% likely to win. IIRC, that’s where they had Obama both in '08 and '12. No polling lead will make me feel even mostly safe if the 538 model doesn’t have the Dem as at least 90% likely to win, and no polling lead will make me feel mostly safe before late October.
iiandyiiii is right - how big a lead it takes is pretty much a function of how far it is from the election. Today, on May 20, I’m not sure any lead would make me feel secure. In September, a large lead would work. In October, I’d feel good with a high-single digits lead, as long as that carries in the battleground states. In November, mid-single digits would work with the same caveat.
Ok, then you need a “Higher than 90%.”
Like, a 10% chance (if we assume polls give an accurate estimate) is still a fucking lot. If I played Russian roulette with a 10 chamber revolver I wouldn’t be all that happy about it.
If it’s 99.9% two minutes before the polls close, I still won’t feel secure. I won’t feel truly secure until Biden takes the oath on Jan. 20, 2021.
The 2016 poll average was more accurate than the 2012 poll average
How close to the election is it? A 10-point lead in May is a lot less meaningful than one in October, especially in a year of big, dramatic news events.
Isn’t that the job of Mike Pence?
However, I totally agree with the basic idea that when your President talks like a dictator, drama wouldn’t end because CNN projects Biden the winner.
I agree. Every day that the election is away adds one percentage point to the uncertainty. Right now, Biden would have to be polling at 230% for me to feel confident.
Thanks for the responses everyone.
Now, for a follow-up question:
What is the minimum polling lead that would have to be overcome for you to suspect that there was cheating at play? For instance, in 2016, nobody seriously claimed Trump cheated to win the election because pre-election polling wasn’t that far off; it was an entirely believable, albeit dismaying, result. But suppose Biden is leading by 10% on the eve of Election Day but Trump ekes out a narrow win, would you suspect cheating? What if 7%?
I will suspect cheating if Nate Silver says that the data suggests that cheating appears to be a likely explanation for some discrepancy.
I distinctly remember when the news called Florida for Al Gore, I turned to my co-worker and said, “Well, Gore won.” Then a while later, they retracted their call. So, as my spouse likes to say while watching football games, “It ain’t over until it’s over.”
And in retrospect Bush v. Gore looks like quaint shenanigans compared to what I’m predicting we will see this coming November (and December and January).
Right.
With most ballots possibly mailed in weeks before, the election eve polling average will not be the only one to consider.
Plus, a polling failure is not evidence of massive cheating. Instead, it is evidence of failed polling.
Silver was very clear that with the paucity of state-level data, there was a 1 in 3 chance of Trump winning. You cannot look at national polls only; the state data is where the value of predictions is best, and gathering data from states was difficult in 2016. I can’t recall, but I think the polling became more accurate in 2018 - it’ll be interesting to see 2020 polls at the state level. Better, I hope.
Trump’s popularity has been badly bruised by his initial response, but he has time to recover. Biden, by contrast, is playing this like a referendum election, which is the smart move, IMO. But I don’t know if he can win just be being option B. If he is to be elected, it will be because Trump’s voters are beginning to peel away.
If you go back to the election of 1932, which in my mind may be the closest parallel to what this election cycle is, FDR was not just an option B; he was offering a combination of a brand name (The Roosevelt name), a lot of optimism, and big plans for the future.
I see Biden as a household name - he’s got that working for him. But he doesn’t necessarily come across as optimistic, and he’s fallen a bit short on offering anything like an FDR-like plan. That’s where he might need to do more than make nice with Bernie Sanders voters - he might actually need to coopt some of his ideas, or at least go halfway in that direction.