Quote Originally Posted by Papewaio View Post
I wouldn't use French politics as the basis. They have a very strong united but small right wing base, and a very broad disjointed left. In the first round it is quite common to have the right wing candidate poll higher than most if not all of the left wing candidates. Then in the final round all the left wing voters rally around a single choice.
Why was the polling gap in 2017 France greater than in all national elections since at least 1969? 10.2% compared to a prior average of 3.9%. The nearest polling gap was in 2002 between Chirac and Le Pen pere. What's the theory here?

And further down in the article, a look at other Euro elections:

Some analysts have argued that people are afraid to admit that they are voting for a far-right candidate such as Le Pen because they don’t want to give a socially undesirable response. That theory was bolstered twice last year, both when the “leave” vote in the U.K. referendum over leaving the European Union did slightly better than polls suggested and when Trump outperformed his polling. But the “shy insert-far-right-candidate here” theory doesn’t hold up when you look at a larger sample of European elections. And it didn’t hold up in France: There was no systematic bias in the polling against the far-right candidate (Le Pen).

As my colleague Nate Silver has pointed out, right-wing populist candidates and parties (in local and parliamentary elections) have, on average, pretty much matched their polling averages3 in European elections since 2012.

Indeed, the French presidential election is the sixth consecutive European election in which the populist right-wing candidate or party underperformed its polling.

None of this is to say that there aren’t “shy voters” in the electorate. It’s just that we may be thinking about them in the wrong way. Instead of undercounting conservative support because people are afraid to give a socially undesirable response, the polls may simply be missing unenthusiastic supporters — people who aren’t excited about their candidate enough to answer a poll but still vote. In fact, when the idea of a “shy” voter was originally formed in 1992, it had nothing to do with right-wing populists. Instead, pollsters were underestimating the strength of the mainstream and relatively milquetoast Conservative Party in the U.K.
[...]
Maybe we should talk less about “shy” voters and more about “apathetic” voters or “reluctant” voters.
What works differently in the USA is that it isn't a two round system nor a preferential voting system. Nor is there compulsory voting either, and by all means it seems stacked with gerrymandering (straight out corrupt in most nations) and making it difficult to vote. So whilst the polls are of a random sample of the voting age people, it doesn't reflect how many are going to actually vote.
Here is the Florida GOP and the federal courts conspiring to suppress the votes of ex-felons, upon whom the former have been striving to impose an unconstitutional poll tax in the form of conditioning voter registration on payment of fines. They've been at it for 1.5 years, since the Florida electorate approved reenfranchisement in a referendum. For extra Kafkaism, Florida does not even maintain proper records on what fines a given applicant owes, with the cherry on top being that the backlog of applications cannot be completed before the 2024 election at current staffing levels. Of course, trying to vote without the satisfaction of the state would be another felony...
https://slate.com/news-and-politics/...-poll-tax.html

You all might recall that I was much more sanguine about the success of reenfranchisement in early 2019.

Process factors beyond the immediate scope of polling, such as Democratic-intending voters going through more trouble voting or getting suppressed by the government, can certainly skew results in theory, assuming pollsters don't try to correct for it. But does it in practice? And to the extent it does, it would have to - in the US - manifest in specific states where these are extant factors. They wouldn't be in all or even most states. You would expect it in Southern and midWestern states in particular. Supporting evidence might be the polling in 2018 that had Dems as mild favorites for the Senate and Gov seats in Florida, but turned into narrow Republican victories. For most recent presidential elections Florida polling has indeed had a Democratic skew. On the other hand, Florida polls - like state polls in general - had a Republican skew in 2012 (when Obama won the state again). This occurred in the context of a national Republican skew in polling that cycle, which implicates internal polling design and not government action.

At all levels, the valence and magnitude of bias change every election, and tends to be associated with identifiable methodological choices that can be adjusted. If one simply assumes a particular static bias, their predictions of electoral outcomes will not be very accurate going by history. Systematic polling bias for left parties has to be measured and analyzed, not just hypothesized.

Australia - election boundaries are set by an independent commission, voting it is on a weekend, postal votes are common, preferential voting is (technically) compulsory with a fine for not attending. The votes still tend to be more right voting than left than indicated in polling. This might be due to the preferential voting and the more extreme right wing groups with a smaller base not getting polled (literally had an idiot Senator from One Nation in on 12 votes - need a massive sample size to capture that in a poll).
The problem is that in the second-party preferred vote the mainstream center-right and center-left parties each typically pull about 50% of the vote. That means polling in aggregate will fall within a single polling error at least. Every election. These are close elections. From what I can see on Wiki Australian polling is precise.


Bottom line is, Biden is multiple polling errors (MOE) ahead of Trump currently, and if that continues to hold until November then going by the entire history of polling a Trump victory would be a vanishingly-unlikely event. The real bias might be the lure of licentious, if often relevant, speculation on tail-end probabilities.

(I will admit that there is clearly an elevated vulnerability in this cycle to unprecedented "rigging" measures, but almost definitionally polling cannot account for lawlessness on such an unprecedented (in our context) scale, so that's kind of beside the point of just how such a hazard might manifest. No poll can weight for a coup, for example. And in practice, as I've been saying, to the extent the electoral process functions semi-normally, sufficient key states to do not have Republicans in charge of the electoral apparatus that the threat remains a mostly-rhetorical one.)