Posted on 08/23/2020 5:05:14 PM PDT by dpwiener
The CBS News national survey was conducted by YouGov using a nationally representative sample of 2,226 U.S. registered voters interviewed between August 19-21, 2020. The margin of error is ± 2.4 points. The CBS News national post-convention survey was conducted by YouGov using a nationally representative sample of 957 U.S. registered voters re-interviewed after the conclusion of the Democratic National Convention, between August 20-22, 2020, and who had been interviewed in CBS News polls prior to the convention. The margin of error is ± 3.6 points.
(Excerpt) Read more at cbsnews.com ...
As you know, we have very different attitudes regarding all of the polls which are out there. You are confident that Biden will win because the poll averages currently have him so far ahead of Trump. Whereas I think most of the polls showing those big leads are either garbage or are deliberately trying to create a false impression so as to boost Biden's chances. The few polls that I consider honest such as Rassmussen, Democracy Institute, and Trafalgar show Trump either leading or roughly tied (within the MOE). That could easily be confirmation bias on my part. But if my viewpoint is correct, it predicts that pollsters with big Biden leads will show those gaps sharply narrowing as we get closer to the election (just like in 2016) so that their final predictions will not be embarrassingly far from the results.
If I'm correct, the question then becomes how the pollsters are skewing their results (leaving aside the issue of whether it's deliberate). The most obvious method is to utilize disproportionate percentages of Democrats, Independents, and Republicans. I decided to do some analysis of the latest CBS Battleground post-DNC-convention poll (August 20-22) that just came out today. (Note that the article is a bit confusing. It shows a pre-convention poll regarding attitudes, taken from 2,226 registered voters, followed by a smaller post-convention poll about actual candidate preferences, taken from 957 registered voters. These are both nationwide polls, not limited to "battleground" states. I'm citing the second poll; just scroll down for the crosstabs.)
The top line indication is that Biden leads Trump 52% to 42%, which appears on the surface to be pretty good for Biden. (Those numbers are rounded; the actual numbers are 51.7% to 42.5% for a 9.1% gap). That compares to the CBS Battleground July 21-24 poll which had Biden leading Trump 51% to 41% (actually 51.0% to 41.1%). Trump appears to have gained 0.8%, which seems statistically insignificant, although it's certainly not the type of historical post-convention bump that Democrats would have been hoping for.
Now let's dig a little further. The latest poll includes 389 Democrats (42.1%), 253 Independents (27.4%), and 282 Republicans (30.5%), making it a D+11.6% poll. That right there would explain the supposed 9.1% gap between Biden and Trump. Even assuming that Democrat voters have an edge on Republican voters, does anyone seriously believe that Democrat turnout in the November election will exceed Republican turnout by 11.6%? Gallup's bimonthly poll of partisan affiliation gives an average in 2020 of 29.7% Democrats, 28.5% Republicans, and 39.1% Independents.
Looking at the details of the CBS July 21-24 poll, it had 542 Democrats (39.4%), 410 Independents (29.8%), and 424 Republicans (30.8%), making it a D+8.6% poll. So even though the latest poll was more biased towards Democrats by 3 additional percentage points, Trump still gained 0.8%. That looks to me like a statistically significant post-convention bump in Trump's favor that the CBS poll was trying to hide by further skewing their results.
The other very interesting aspect of these two polls is that Independents are under-represented (according to Gallup), and that effect was exacerbated in August, going from 29.8% in July down to 27.4%. In July, Trump led Biden by 43% to 40% among Independents. But in August that 3% lead jumped up to 10% (47% to 37%) among Independents.
To summarize, the internals of these CBS Battleground polls tell me the following:
* A more plausible weighting of sample percentages for Democrats, Republicans, and Independents would show Biden and Trump in a virtual tie, or perhaps a slight Trump lead.
* Trump received a post-DNC-convention bump.
* Trump's lead among Independents is growing and is substantial.
* Biden is in trouble even when focusing on the biased polls, and it's going to get a lot worse for him as the gaps narrow approaching the election.
Do they really not adjust for the large dem size?
I find that hard to believe.
Wow.
I recall in 1988 when they were saying how wonderful Olympia Dukakis’ relative was doing (for his state and the polls - The Massachusetts Miracle!)
SeeBS has been lying to the public for a very long time
Your corrections would match Rasmussen’s survey.
Between the sampling distortions favoring Biden, the silent DjT voters, the EV in key swing states, and a greater black vote (could already be baked into the silent vote) I’m still expecting a drubbing for the Ds. Larger than ‘16’s 306EV, I think it was.
The poll says that "This sample was weighted according to gender, age, race, and education. ... Respondents were selected from YouGov's opt-in panel to be representative of registered voters in the U.S."
It says nothing about partisan weighting, which they apparently did not consider relevant.
I agree with you completely. That's why I made the bet with my brother, even though I'm not a Trump voter. (I'll be voting for Jo Jorgensen. Of course my Presidential vote won't matter, since I live in California.)
I agree they are biased. A blow out—for either side—will affect the audience. None of these outlets can afford to lose audience.
I think if Trump is even among people who are actually willing to talk to the very few honest pollsters, he is fifteen points ahead among likely voters. Given the unusually high cheat factor this year, subtract a maximum of 10% from that - for an actual lead of five points.
If your vote did matter, would you still throw it away by voting for Jo Jorgensen?
Trump will have to use the power of Federal law enforcement in some way to counteract that. He has been reluctant to do so up until now - we can hope it will be his October Surprise.
Do you have any feeling how Newsome and that AB5 gig legislation may affect local and district races ?
I expect some R losses in ‘18 to be recovered. More than some.
Accurate polling today unlike just 20 years ago, is a near impossible task. The first problem is how do you get a truly random sample. A large segment of the population do not have a landline and do not answer calls from numbers they don’t recognize. This adds a bias to the sample that is hard to correct for. Web based polls are even worse. Then how do you correct for voter enthusiasm and motivation? Finally, you have the Trump Effect voters. These are working class voters who have never voted for anyone except for Trump, because neither party has represented their interests. Trump not only speaks to their concerns he has followed through on his promises and consequently they will come out to vote for him. This group may be 500,000 to 1,000,000 strong.
essentially, Basementman and Willie’s Mistress are sitting out the campaign season. They are just sitting back and letting the media do the campaigning for them. They are betting that the majority of people hate Trump as much as they do and that all they have to do is sit back and get anointed. Most people expect those wanting to be president to actually put some effort into getting their vote, usually by coming to their state and showing some interest. Every poll I’ve seen show that there is no real enthusiasm for Basementman, but there is a lot of enthusiasm for Trump.
My bet is that Trump will win the EC and the popular vote this time.
The other aspect of this poll is weighting by education. This was a huge miss by pollsters in 2016 when they over-estimated college educated fraction of voters. I do not see any education-related adjustment in the CBS poll.
Trump knows this and is on the case, if every any candidate was on a case.. Attempts will be made to beat DjT because Ds believe everyone hates him, so no one will mind if they cheat in favor of the “good guys.” Extremely dangerous attitudes.
Trump knows. Counter actions are in place.
We know law enforcement will be on call if poll intimidation pops up. I expect legal threats if there’s discrepancies like those just exposed in Michigan.
Some of the states may get away with ballot harvesting, (which will be legislated as illegal in the new R Congress), but will have limited affect now that we’re wise to the scam. Little worry, I hope.
I do worry about some swing districts, as in California, that were stolen via ballot harvesting in ‘18, but again, we’ve been forewarned. I assume Rs are planning on ballot harvesting, too, where legal.
“Of course my Presidential vote won’t matter...”
Then why vote?
BS, PDJT is up by about 15. Sampling cannot be even proven to have occurred. All they need to do is project the numbers they want. Find me a conservative who answers a poll. Broken glass election
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.