The Quinnipiac poll showing Warren in the lead sampled self identified registered voters not likely

[Demsrule86]

voters. This explains this outlier...but even here most expected that Biden would beat Trump most easily....

"This survey uses statistical weighting procedures to account for deviations in the survey
sample from known population characteristics, which helps correct for differential survey
participation and random variation in samples. The overall adult sample is weighted to
recent Census data using a sample balancing procedure to match the demographic makeup of
the population by region, gender, age, education and race."

The polls are the polls of course but using self identified registered voters ...random calling and statistical methods which I think are questionable call into question the accuracy of this poll . The folks they call claim to be Democrats and maybe most are but who knows...Since this poll is an outlier, it is my opinion that the methodology explains why this is so...and we shall see but I doubt its veracity.

I would also add the internals are not very well defined. Democrats from lists of Democratic voters and likely voters should be used particularly in a primary for polling and not random calling which ended using self identified registered Democrat and 'Democratic leaning voters'. Next the data was massaged using two methods which I find disturbing...but IMHO, the voters called did not produce and adequate sampling of Democratic voters and the way the data was manipulated disturbed me.

https://poll.qu.edu/national/release-detail?ReleaseID=3646
https://poll.qu.edu/images/polling/us/us10242019_demos_ulow46.pdf/

[Loki Liesmith]

1. I find it unhelpful to spend too much time on internals

The poll said what it said. Throw it in the pile and work with all the other numbers.

[Demsrule86]

2. The data was manipulated and they sampled self identified registered voters...I just think this is a

poll where someone has their thumb on the scale.

[bluewater]

4. Um... SUSA POll: question 1. Are you registerd to vote?

Um... but the Survey USA Poll that was good for Biden, and coincidentally Biden supporters claim it is one of the best polls, very FIRST question is ARE YOU REGISTERED TO VOTE?

They then, in there breakdowns of candidate preferences, refer to the respondents as "likely Democratic Primary voters".



Quinnipiac is a respected major poll, and is rated A- by 538, but feel free to "unskew" the poll because you feel it is an "outlier".

Me, I will go with 538's assessment of Quinnipiac and their methodology.

Honestly, saying "the data was manipulated" in the Quinnipiac Poll is going to require proof which you did not provide.

[Demsrule86]

8. I don't favor polls that use random sampling and self -identified Democrats for primaries...

However, I do not believe they used Democrat leaning voters in the poll that had Biden leading...so this one may be a bit worse...but I don't like the one you mention either...for the reasons I have stated...for primaries, you need to sample likely voters from Democratic voter lists...

[bluewater]

10. But you didn't say that about the SUSA poll when it came out...

You seemed to have focused on the poll that was unfavorable for Biden.

[Celerity]

12. if people like the poll result, it is gold, if they do not, it is a brick of shit nt

[Skya Rhen]

3. Thank you for explaining the workings of this poll. I don't give Quinnipiac much credence.

[ritapria]

5. Qhinnipiac is wildly overrated by the media

Survey USA is wildly underrated - completely ignored by the media

[bluewater]

6. I think the media likes the Quinnipiac Poll becuase it is 100% random live callers.

The Survey USA poll is not, it's partly online, partly live caller.

Both are good polls, 538 rates Quinnipiac A-, and SUSA A.

[Demsrule86]

7. We will see...I don't think this is really a great poll...as I states...self identified Democratic

Democratic leaning voters? Also, the data was weighted and massaged.

[bluewater]

11. "Also, the data was weighted and massaged." lol

People better HOPE that random live caller polls, heck ANY poll, weight their results by things like age, education, ethnicity and income levels to match national averages.

All good polls do that.

Other wise you can get meaningless polls stacked with one demographic thru random chance.

Pardon me for asking, but how do you think polls work? That people are called up randomly and those results go straight into the published poll results with no "massaging" for any sampling bias?

I am stunned.

[Hav]

9. I think too much effort is put into trying to find fault

with polls one doesn't like. It's done here whether it's good polls for Biden or good ones for Warren.

Something that I think is a legitimate point is the distinction of registered vs. likely voters.
Otherwise, using statistical methods and using weighting so that your sample is in line with what you consider a representation of the population is the bread and butter of pollsters. There is nothing wrong about it, it's not unusual and I'd expect most pollsters to do that.