The USA Today/Suffolk poll showing Trump beating all Dem candidates had a serious flaw

[highplainsdem]

that explains why it's an outlier.

The pollster allowed voters to say whether they'd prefer to vote for an unnamed candidate running as the nominee of an unnamed third party. In other words, the pollster gave the voters almost unlimited choices besides Trump versus each Democratic candidate. That third option got 11-15% of the vote, skewing the results in Trump's favor...and that high percentage for an unnamed third party candidate was higher than in any election since 1992, when Ross Perot ran, so it's unlikely to be an accurate prediction of how well third party candidates would do next year.

I hadn't read much of the article about the poll

https://www.usatoday.com/story/news/politics/elections/2019/12/16/trump-impeachment-2020-election-leads-democratic-rivals/2663659001/

and didn't notice what it said about this odd aspect of the poll:

President Donald Trump, the first modern president to face impeachment during his first term in the White House, now leads his top Democratic rivals in his bid for a second, a new USA TODAY/Suffolk University Poll finds.

The national survey, taken as the House of Representatives planned an impeachment vote and the Senate a trial, showed Trump defeating former Vice President Joe Biden by 3 percentage points, Vermont Sen. Bernie Sanders by 5 points, and Massachusetts Sen. Elizabeth Warren by 8 points.

In hypothetical head-to-head contests, Trump also led South Bend, Indiana, Mayor Pete Buttigieg by 10 points and former New York City Mayor Michael Bloomberg by 9.

-snip-

Trump's standing remained remarkably steady regardless of his opponent, at 45% against Warren, 44% against Biden and Sanders, and 43% against Buttigieg and Bloomberg. That could be both good news and bad for him: A sign of the solidity of his support, but also an indication that he has a ceiling.

There was more variation among the Democratic contenders when they were matched against Trump. Biden received 41%, Sanders 39%, Warren 37%, Bloomberg 34% and Buttigieg 33%.

An unnamed third-party candidate received between 11% and 15% in the head-to-head contests – a factor that could determine who wins the White House.

Emphasis added.


My attention was called to this flaw in the poll by a Nate Silver thread about electability which averaged four recent national GE polls including the Suffolk poll

Link to tweet

and which got this response from John Barro:

Link to tweet

What do you make of the Suffolk format where they prompt the option of voting for "a third party candidate"? Result is higher than we've seen for third party candidates in any election since 1992, which doesn't sound right.

to which Nate replied:

Link to tweet

I'm generally not a fan of any polling that allowes for generic/unnamed candidates

Btw, that Nate Silver thread, which starts with the tweet below, makes the point that averaging recent GE polls, which have become tighter, shows Biden beating Trump by 4.5 points, Sanders beating him by 2, but Warren and Buttigieg both losing to Trump, Warren by 1.2 and Buttigieg by 1.5.

As Nate says, this "may raise the salience of electability/beating Trump."

[LeftTurn3623]

1. Yikes - I might have to support Biden after all

[Fiendish Thingy]

2. If Nate acknowledges the flaws in the Suffolk poll, then why include it in the average? Nt

[warmfeet]

3. All I know is that I

will be voting for the Democratic nominee. Warren, Sanders, Biden, Buttigieg - whoever.

[Cha]

4. Good to know.. Mahalo, hpd!