US Politics Mega-thread - Page 2736

On October 11 2020 23:35 JimmiC wrote:


They are mainly picking apart your first paragraph, being as though all probability is a mathematical derived guess. And you could get a fair bit of comfort that it was better than a coin flip by checking its performance in all elections not just the one.

https://projects.fivethirtyeight.com/checking-our-work/

https://www.google.ca/amp/s/fivethirtyeight.com/features/how-fivethirtyeights-2018-midterm-forecasts-did/amp/

https://fivethirtyeight.com/features/why-fivethirtyeight-gave-trump-a-better-chance-than-almost-anyone-else/

On October 11 2020 23:44 Simberto wrote:
There is really no need to turn this into a major mudfight. GH said something, people responded, GH clarified, stuff is clear now. No need to fight about who won that exchange.

On October 11 2020 23:06 GreenHorizons wrote:

28.6% or 15% ≠ 35%? Are you guys messin with me? I know (granted not nearly as well as Simberto) how probability works.

I was clarifying that saying "Trump's chance to win was 35% on election day" was flat wrong because that's not what 538 (where the 2020 probability was referenced from) actually said. It said 28.6% (or 28.2% for "polls plus" since it's unclear which, is being used in 2020) and NYT had him at 15% on election day.



Also this ^

and: since the "35% chance on election day" was being used to support an argument that his chances today in 2020 are significantly worse than this time in 2016.

On October 12 2020 00:21 GreenHorizons wrote:

Indeed. I simply wanted to clarify that:

1. The "35% chance to win on election day" was incorrect it was between 15% (NYT) and 28.6% (538)
2. The polling data used to calculate 538's "chance to win" is giving approximately the same results as it did this time last Presidential election* despite the differences recently mentioned
3. "The problem with these chances is that they are hard to actually test, because you can not have 1000 elections from the starting position" so we can't really have much clue as to whether it has any more accuracy than assigning a random chance to win to both candidates.

This is unlike rolling a die, flipping a coin, or drawing a pocket pair in hold 'em where we can all try it ourselves and check to see if you're right that there is a ~1/6 chance of rolling a 3 or a ~50% chance of a coin falling heads up, or ~6% chance at a pocket pair. Presumably the more trials we do, the closer our experimental probabilities will get to the theoretical unless there's an unaccounted for variable reliably distorting the results (loaded dice for example).

On October 12 2020 00:35 LegalLord wrote:

As I noted in a write-up from last election, it's less so "if you did this 1000 times" and more so "a degree of belief that this is the outcome suggested by the data." Potentially a subtle point, but one that makes a difference when the event happens only once.

The differences from last time that makes a Biden win that much more probable is that:

1. Polls haven't really been very volatile; the needle hasn't moved much since March.
2. There are a lot fewer undecided voters than there were in 2016.

So it might not be a surprise that 538, for example, would give Biden a better chance of victory than last time. As of right now I see them giving an 86% probability of winning.

On October 12 2020 01:12 GreenHorizons wrote:


Just to be clear, Clinton's chance to win this date in 2016 was 83.5% and rose to 86.9% on the 12th. If 538 thinks the data reflects a better chance for Biden to win this election than Hillary last election, it doesn't seem to show up in their probability calculation.

That's really what I was curious about. If there's reliable data that shows Biden is notably more likely to win than Clinton, why wouldn't 538 be giving him a significantly better probability of winning compared to her at this time last year?

On October 11 2020 18:09 GreenHorizons wrote:


NYT had him at 15% and 538 had him at 28.6% chance of winning on election day.

538 had him at about 13-14% ~this week of October

NYT had him as low as 7% in mid-late October

www.nytimes.com

projects.fivethirtyeight.com

EDIT: It makes me wonder how 538 is coming to essentially the same conclusion despite this difference in polling/favorability between 2016 and 2020.

On October 12 2020 01:23 Nevuk wrote:

538 is factoring in instability from COVID and a little bit from Trump's mail-in ballot claims/voter suppression - so they treat the possibility of very unusual outcomes as higher than in 2016 (ie huge polling misses). They just put out a podcast about how the model is reacting to the massive leads that I have yet to listen to, but I imagine it addresses this slightly.

538 also had 3 versions of their model last year, and all of them said similar but slightly different things (the fundamentals of the race favored the challenger somewhat, so the more non-polling data used the better Trump did).

On October 12 2020 00:35 LegalLord wrote:

As I noted in a write-up from last election, it's less so "if you did this 1000 times" and more so "a degree of belief that this is the outcome suggested by the data." Potentially a subtle point, but one that makes a difference when the event happens only once.

The differences from last time that makes a Biden win that much more probable is that:

1. Polls haven't really been very volatile; the needle hasn't moved much since March.
2. There are a lot fewer undecided voters than there were in 2016.

So it might not be a surprise that 538, for example, would give Biden a better chance of victory than last time. As of right now I see them giving an 86% probability of winning.

On October 12 2020 01:31 GreenHorizons wrote:


I don't know what "factoring instability from COVID and mail in ballot claims" means from a mathematical/statistical point of view but those are certainly variables that would have changed since 2016 and I can imagine turning a better polling position by Biden into basically the same chance to win as Clinton had with much less favorable polling (her polling well under 50% and net unfavorable) at this point in the race in 2016.

Wish you could have been the first response

In forecasting how much the polls could change, we now account for more components related to uncertainty. Two of these components include estimating i) economic uncertainty and ii) the overall volume of important news, both of which are very high under COVID-19. These offset other trends — such as greater polarization — that would lead to less uncertainty.
We’ve put a lot more work into our economic index: i) extending it back to 1880 to capture a fuller range of economic conditions, ii) adjusting it for increased partisanship and iii) developing an economic forecasting component to reflect potential changes in the economy between now and November. This is important because most projections forecast substantial improvement in the economy before November.
We attempt to account for additional uncertainty in Election Day results because turnout will potentially be less predictable given the pandemic.
We allow COVID-19 to be a factor in determining covariance. That is to say, states that have had high rates of COVID deaths and cases (such as Arizona and New York, which otherwise don’t have that much in common politically) could have correlated outcomes. Likewise, we also consider covariance based on a state’s anticipated rate of mail voting.
With the party conventions being substantially scaled down and largely held virtually, we’re applying only half of the usual “convention bounce adjustment” (see below for more on the convention bounce adjustment).
[... non COVID changes]
We now account for changes in how easy it is to vote in each state, as empirically, this yields higher turnout and a higher share of Democratic votes.

The components of our uncertainty index are as follows:

The number of undecided voters in national polls. More undecided voters means more uncertainty.
The number of undecided plus third-party voters in national polls. More third-party voters means more uncertainty.
Polarization, as measured elsewhere in the model, is based on how far apart the parties are in roll call votes cast in the U.S. House. More polarization means less uncertainty since there are fewer swing voters.
The volatility of the national polling average. Volatility tends to predict itself, so a stable polling average tends to remain stable.
The overall volume of national polling. More polling means less uncertainty.
The magnitude of the difference between the polling-based national snapshot and the fundamentals forecast. A wider gap means more uncertainty.
The standard deviation of the component variables used in the FiveThirtyEight economic index. More economic volatility means more overall uncertainty in the forecast.
The volume of major news, as measured by the number of full-width New York Times headlines in the past 500 days, with more recent days weighted more heavily. More news means more uncertainty.

In 2020, measures No. 1 through 5 all imply below-average uncertainty. There aren’t many undecided voters, there are no major third-party candidates, polarization has been high and polls have been stable. Measure No. 6 suggests average uncertainty. But metrics No. 7 and 8 imply extremely high uncertainty; there has been a ton of news related to COVID-19 and other major stories, like the protests advocating for police reform in response to the death of George Floyd — not to mention the impeachment trial of Trump earlier this year. Likewise, there has been as much volatility in economic data as at any time since the Great Depression.

On the one hand, the sheer number of uncertainties unique to 2020 indicate the possibility of a volatile election, but on the other hand, there are also a number of measures that signal lower uncertainty, like a very stable polling average. So when we calculate the overall degree of uncertainty for 2020, our model’s best guess is that it is about average relative to elections since 1972. That average, of course, includes a number of volatile elections such as 1980, 1988 and 1992, where there were huge swings in the polls over the final few months of the campaign, along with elections such as 2004 and 2012 where polls were pretty stable. As voters consume even more economic- and pandemic-related news — and then experience events like the conventions and the debates — it’s not yet clear whether the polls will remain stable or begin to swing around more.

On October 12 2020 01:54 Mohdoo wrote:
Biden not having a 20 year cultivated culture of hatred against him certainly lowers the % of undecided voters. So many people saw Trump and Clinton as untenable. Biden is totally fine for a lot more people. That and I think a lot of people are finding themselves saying "ok, Trump is having legitimately bad consequences, such as covid"

On October 12 2020 01:54 Mohdoo wrote:
Biden not having a 20 year cultivated culture of hatred against him certainly lowers the % of undecided voters. So many people saw Trump and Clinton as untenable. Biden is totally fine for a lot more people. That and I think a lot of people are finding themselves saying "ok, Trump is having legitimately bad consequences, such as covid"

On October 12 2020 02:13 Blitzkrieg0 wrote:


Every attack ad I see from Trump's campaign is how Biden is a radical leftist. I doubt that has as much effect on younger people, but boomers have been trained to hate socialism for their entire lives.

On October 12 2020 02:13 Blitzkrieg0 wrote:


Every attack ad I see from Trump's campaign is how Biden is a radical leftist. I doubt that has as much effect on younger people, but boomers have been trained to hate socialism for their entire lives.