Yahoo News/YouGov: Biden +6

[Amenhotep Bakari-Sellers]

As I stated earlier, Trump's quick response to Laura, has earned him pts with soccer moms, eventhough protesters in WI are trying to get Trump out of office. Soccer moms swing back and forth as Larry Sabato said, that's why we see huge discrepancies in the polls due to the protests whom are trying to get Trump out and the issues like the economy,  which has unemployment dropping to 7.5 percent by Nov

That's why Biden is visiting MN, WI and PA next week

[Interlocutor is just not there yet]

Godawful poll for Biden considering this is a poll of US Adults. Trump always does better in RV and especially LV screens so this implies the race is actually competitive if the election happens during an upswing for Trump. I'd expect this poll to bounce back to Biden +9 or so in a week or two once the bounce wears off.

Current projection: Biden +4, Biden 280-Trump 260.

No it's not. The 47-41 result is based off of registered voters

Could you share your source? It says ‘1,001 US Adults’ at the top of all of the opening slides, including the one with the 47-41 result.

The link in the first post?

The header looks like this on page 17:

"4. 2020 Election - Presidential Vote Intention
Who will you vote for in the election for President in November?
Among registered voters"


The Unweighted respondents is listed as 807. The crosstabs also have a 'Registered Voters' subgroup with the exact same numbers.

They interviewed 1,001 adults in total, but not everyone had the same questions. For example, they asked the "Is your mind made up?" question only to Biden/Trump voters (710 of them)

[Fmr. Gov. NickG]

It looks like the unweighted result of the horserace Q was around Biden +11.

That’s not to say they should have published a Biden +11 topline.  There are almost certainly too many Dems in the sample just based on the 2016 reported vote.

But it’s does demonstrate how the results of these online polls are often much more a function of weighting decisions by the pollster than any sincere attempt to get a representative sample in the first place.

This wouldn’t necessarily create any systematic bias in the poll’s reported result (unlike what Trafalgar is doing, for instance).  But it will often exaggerate any idiosyncracies in response from the part of the population that got undersampled. (Most famously in the case of the one young black Trump supporter that got counted as like 30 people in each iteration of the USC poll last cycle.)

[Umengus]

It looks like the unweighted result of the horserace Q was around Biden +11.

That’s not to say they should have published a Biden +11 topline.  There are almost certainly too many Dems in the sample just based on the 2016 reported vote.

But it’s does demonstrate how the results of these online polls are often much more a function of weighting decisions by the pollster than any sincere attempt to get a representative sample in the first place.

This wouldn’t necessarily create any systematic bias in the poll’s reported result (unlike what Trafalgar is doing, for instance).  But it will often exaggerate any idiosyncracies in response from the part of the population that got undersampled. (Most famously in the case of the one young black Trump supporter that got counted as like 30 people in each iteration of the USC poll last cycle.)

The reason of "weighting decisions by the pollster" is to have a representative sample. And you are right, the way of weighting explains generally the result.

But lots of people forget that we are only in august and the majority of pollsters uses RV again. Polls will be relevant when LV screen will be used. IMO, it will boost Trump.

[MillennialModerate]

This puts a pit in my stomach