- My Forums
- Tiger Rant
- LSU Recruiting
- SEC Rant
- Saints Talk
- Pelicans Talk
- More Sports Board
- Fantasy Sports
- Golf Board
- Soccer Board
- O-T Lounge
- Tech Board
- Home/Garden Board
- Outdoor Board
- Health/Fitness Board
- Movie/TV Board
- Book Board
- Music Board
- Political Talk
- Money Talk
- Fark Board
- Gaming Board
- Travel Board
- Food/Drink Board
- Ticket Exchange
- TD Help Board
Customize My Forums- View All Forums
- Show Left Links
- Topic Sort Options
- Trending Topics
- Recent Topics
- Active Topics
Started By
Message
Analysis of Polls vs. Actual Results
Posted on 9/21/20 at 10:42 am
Posted on 9/21/20 at 10:42 am
So I went back and looked at the final RCP average battleground state polls from 2016 and compared them to the actual results to see if I could find a "Hidden Trump Voter" factor to apply to this year's polls.
Pennsylvania was where the polls were way off. Clinton was +6.2 on Election Day. Trump won by 0.7 (? 6.9)
Florida was 0.0, Trump won by 1.2 (? 1.2)
Wisconsin was Clinton 4.0, Trump won 0.7 (? 4.7)
Michigan was Clinton 5.2, Trump won 0.3 (? 5.5)
North Carolina was Trump 1.7, Trump won by 3.7 (? 2.0)
In these five states, the ? from poll result to actual results averaged 4.09 in favor of Trump.
This year, these states' polls look like this.
PA - Biden 4.0
FL - Biden 1.6
WI - Biden 6.7
MI - Biden 4.8
NC - Biden 0.9
If you apply the mean ? to these five states, then Trump would win FL, PA and NC, which would leave him one state short of winning. I believe that this is how the Democrats are interpreting the polls.
I believe the better format is apply the median ? of 4.7, which would give Trump PA and make Michigan a statistical tie.
In short, add between 4.09 and 4.7 to Trump's numbers in any battleground state poll to see what the 2016 electorate applied would look like.
I also do not believe that Democrats will come out like they did for Hillary. I believe that there were 3.8 million women who voted for Hillary only because she could have made history as the first woman president. Some will come with Kamala on the ticket, but not as many. This is what is putting states like Minnesota in play. Lots of feminists in Minnesota who aren't excited about voting for an old white man.
Edit: The ? above is supposed to be the symbol for Delta. Unicode not allowed on TD.
Pennsylvania was where the polls were way off. Clinton was +6.2 on Election Day. Trump won by 0.7 (? 6.9)
Florida was 0.0, Trump won by 1.2 (? 1.2)
Wisconsin was Clinton 4.0, Trump won 0.7 (? 4.7)
Michigan was Clinton 5.2, Trump won 0.3 (? 5.5)
North Carolina was Trump 1.7, Trump won by 3.7 (? 2.0)
In these five states, the ? from poll result to actual results averaged 4.09 in favor of Trump.
This year, these states' polls look like this.
PA - Biden 4.0
FL - Biden 1.6
WI - Biden 6.7
MI - Biden 4.8
NC - Biden 0.9
If you apply the mean ? to these five states, then Trump would win FL, PA and NC, which would leave him one state short of winning. I believe that this is how the Democrats are interpreting the polls.
I believe the better format is apply the median ? of 4.7, which would give Trump PA and make Michigan a statistical tie.
In short, add between 4.09 and 4.7 to Trump's numbers in any battleground state poll to see what the 2016 electorate applied would look like.
I also do not believe that Democrats will come out like they did for Hillary. I believe that there were 3.8 million women who voted for Hillary only because she could have made history as the first woman president. Some will come with Kamala on the ticket, but not as many. This is what is putting states like Minnesota in play. Lots of feminists in Minnesota who aren't excited about voting for an old white man.
Edit: The ? above is supposed to be the symbol for Delta. Unicode not allowed on TD.
This post was edited on 9/21/20 at 10:43 am
13
Posted on 9/21/20 at 10:45 am to anc
according to polls, the errors that were made in 2016 have been corrected
Posted on 9/21/20 at 10:46 am to anc
quote:
"Hidden Trump Voter" factor
Is bigger this year...substantially bigger
Posted on 9/21/20 at 10:46 am to anc
Eh, women hate whores. I doubt Kamala gets the women vote at least suburban women.
Posted on 9/21/20 at 10:48 am to ksayetiger
quote:
according to polls, the errors that were made in 2016 have been corrected
Yep, I would go with the word of someone that far off last time.
Posted on 9/21/20 at 10:48 am to ksayetiger
quote:
according to polls, the errors that were made in 2016 have been corrected
No they haven't. The People's Pundit guy said that the polls are actively trying to factor in mail in voting so their likely voter screens, if they are using them at all, are way too generous. Some polls are still using registered voters because they think mail in voting will mean there is no point in using likely voter screens since every registered voter will be able to vote.
I got polled yesterday for the first time. From Public Policy Polling, a known democrat affiliated pollster. They asked me who I planned on voting for in the first question, then asked me a bunch of leading questions that had to do with the supreme court (i.e "If you knew that your choice for President would appoint a supreme court justice who would take away a woman's right to choose, would it make you more likely or less likely to vote for them?") and then the final question was asking me again who I planned on voting for.
I bet they used the results of the 2nd time they asked the question for their topline after asking a bunch of leading questions.
Posted on 9/21/20 at 10:49 am to anc
Why not just apply the error rate from 2016 directly to the State in question. That would result in
PA - Trump 2.9
FL - Biden 0.4
WI - Biden 2.0
MI - Trump 0.7
NC - Trump 1.1
PA - Trump 2.9
FL - Biden 0.4
WI - Biden 2.0
MI - Trump 0.7
NC - Trump 1.1
Posted on 9/21/20 at 10:51 am to anc
quote:
also do not believe that Democrats will come out like they did for Hillary.
Great analysis and it shows that the battleground states will be razor thin.
Do you have data to support the statement above? I think one could also make a good argument that Democrats stayed home in 2016 due to erroneous polls, and will not make that mistake in 2020.
Posted on 9/21/20 at 10:51 am to anc
Yet, how can you factor in the unadulterated fraud inherent in those states with rampant mail in voting?
I dont think anyone can statistically account for a variable we dont truly understand, only that it will have some skew towards D.
I dont think anyone can statistically account for a variable we dont truly understand, only that it will have some skew towards D.
Posted on 9/21/20 at 10:52 am to RICHIE APRILE
quote:
(i.e "If you knew that your choice for President would appoint a supreme court justice who would take away a woman's right to choose, would it make you more likely or less likely to vote for them?")
That is a horrible polling question. Any student who has taken my polling analytics class would have failed for this. Its literally the first point my textbook makes!!!!!!!
Really strange that I am working on a lecture on this. Thanks for the heads up!
Posted on 9/21/20 at 10:52 am to ksayetiger
quote:
according to polls, the errors that were made in 2016 have been corrected
How do you correct for people not answering your calls or lying to you? The same environment, probably more extreme, exists today.
Posted on 9/21/20 at 10:52 am to Weekend Warrior79
quote:
according to polls, the errors that were made in 2016 have been corrected
According to the Congressional Presidential Election Committee of 2016 and 2018...That statement is FOS.
Posted on 9/21/20 at 10:52 am to anc
I think you have it backwards I think the Dems come out more for Biden. Hillary was that disliked
Posted on 9/21/20 at 10:53 am to ksayetiger
How does this post not have multiple upvotes? Too subtle.
This post was edited on 9/21/20 at 10:54 am
Posted on 9/21/20 at 10:54 am to anc
quote:
I also do not believe that Democrats will come out like they did for Hillary.
Apparently the city of Detroit is an excellent indicator of election day turnout.
I knew a DNC strategist when I lived in DC during the last election. He said he knew Trump was going to win when he got a report that lines in Detroit's most populated polling places were nearly non-existent at 4pm on election day.
Of course, it is going to be hard to predict voter turnout with mail-in voting, but we should have a good idea of who will win by mid-afternoon on November 3rd based on the turnout in Detroit.
Posted on 9/21/20 at 10:57 am to Bluefin
quote:
Apparently the city of Detroit is an excellent indicator of election day turnout.
I knew a DNC strategist when I lived in DC during the last election. He said he knew Trump was going to win when he got a report that lines in Detroit's most populated polling places were nearly non-existent at 4pm on election day.
Of course, it is going to be hard to predict voter turnout with mail-in voting, but we should have a good idea of who will win by mid-afternoon on November 3rd based on the turnout in Detroit.
People's Pundit guy said Trump won Michigan only because the turnout for Hillary was terrible. He said every other state Trump won, Trump simply won by beating Hillary straight up, except Michigan, where it was the lack of turnout on Hillary's end. I don't see them being caught flat footed this time.
Posted on 9/21/20 at 10:59 am to anc
I still don’t understand the reason for the consistent hidden 2016 Trump poll numbers. Can someone here explain that?
Posted on 9/21/20 at 11:00 am to anc
quote:
quote:
(i.e "If you knew that your choice for President would appoint a supreme court justice who would take away a woman's right to choose, would it make you more likely or less likely to vote for them?")
That is a horrible polling question. Any student who has taken my polling analytics class would have failed for this. Its literally the first point my textbook makes!!!!!!!
As a generic question, yes, I'd agree. As a question designed to push people out of some comfort zone of poll response mode, or, to see if they flip flop on how they answer given different stimuli, I'd say its fair. This would also need to be a question toward the end of the survey. Consultants are crafting ad campaigns off this research, so I would want to know how to possibly flip a voter.
Another example, would a luke warm Biden supporter maybe rethink their preference if you asked the same but said they'd take away the 2nd Amendment.
Posted on 9/21/20 at 11:03 am to SlickRick55
quote:
I still don’t understand the reason for the consistent hidden 2016 Trump poll numbers. Can someone here explain that?
Because a lot of people are hesitant to trust pollsters, especially live interviewers.
Posted on 9/21/20 at 11:05 am to anc
Biden landslides any state allowing mail-in voting. Thats a fact no way around it.
Page 1 of 2
Back to top
Follow TigerDroppings for LSU Football News