New poll hype thread

[amdcpus]

CBS/YouGov has two state polls coming out tomorrow, from Nevada and Utah. I'll guess:

NV: Clinton +4 , Cortez-Masto +2
UT: Trump 31 - Clinton 27 - McMuffin 24

I doubt McMullin will be that high. They don't include third party candidates in either the question or answer. You have to select "someone else" and type their name in.

[amdcpus]

Following amdcpus's lead, I currently have a one-question Google Consumer Survey South Dakota out in the field.  

The question I asked was: If the presidential election were held today, who would you vote for?
The choices are: (randomized) Donald Trump of the Republican Party, Hillary Clinton of the Democratic Party, Gary Johnson of the Libertarian Party, Darrell Castle of the Constitution Party and "I am not registered to vote in South Dakota" (always last).  According to Balletopedia, those are the only 4 candidates actually on the ballot in SD.  Stein didn't make the cut.  

I added the not registered to vote option to make it more like a RV poll (of course, it's technically not, and it's impossible to ferret out RV/LV when you only get one question) and to try to weed out non-South Dakotans with a South Dakota IP address.  I didn't include Undecided because, as I understand it, those who are can skip the question in GCS, anyway.

I have to figure out how to weigh the sample by age.  Google weighs by age of the Internet population, which skews younger than RVs.  There was no 2012 exit poll for SD, so I have to use something else (probably Census estimates of 2012 turnout or RVs).  Any thoughts on that would be appreciated.

I have an excel sheet I made to weight the numbers and a turnout by age doc I can give you.

[amdcpus]

That's actually pretty cool. How much did that cost you?

$75 for 500 respondents.  (i.e. 15 cents per respondent)  The number of pseudo "registered" voters in the poll will be less than that because I'll have to back out the self-reported not SD registereds.

I have an excel sheet I made to weight the numbers and a turnout by age doc I can give you.

Is it the same one you posted in the Utah poll thread?  What benchmark did you use?  Do you have something SD-specific?  

I am leaning toward weighting according to the number of 2012 RV in each age class, as reported by the U.S. Census Bureau.

I'll post a link to it later, but it does have the data for SD as well.

[amdcpus]

Following amdcpus's lead, I currently have a one-question Google Consumer Survey South Dakota out in the field.  

The question I asked was: If the presidential election were held today, who would you vote for?
The choices are: (randomized) Donald Trump of the Republican Party, Hillary Clinton of the Democratic Party, Gary Johnson of the Libertarian Party, Darrell Castle of the Constitution Party and "I am not registered to vote in South Dakota" (always last).  According to Balletopedia, those are the only 4 candidates actually on the ballot in SD.  Stein didn't make the cut.  

I added the not registered to vote option to make it more like a RV poll (of course, it's technically not, and it's impossible to ferret out RV/LV when you only get one question) and to try to weed out non-South Dakotans with a South Dakota IP address.  I didn't include Undecided because, as I understand it, those who are can skip the question in GCS, anyway.

I have to figure out how to weigh the sample by age.  Google weighs by age of the Internet population, which skews younger than RVs.  There was no 2012 exit poll for SD, so I have to use something else (probably Census estimates of 2012 turnout or RVs).  Any thoughts on that would be appreciated.

That's actually pretty cool. How much did that cost you?

I have an excel sheet I made to weight the numbers and a turnout by age doc I can give you.

Do you work for a polling firm or something? You don't have to say which, but I'm curious.

Nope, just enjoy math.

[amdcpus]

Template: https://docs.google.com/spreadsheets/d/10semu5xRZ-D6gG7J3d19asXVMnoCvh7G2RNuySOv_RI/edit?usp=sharing

Template with my Utah Poll: https://docs.google.com/spreadsheets/d/1bWqN7yncLyttlivU1hc9FX7xN8jZ4o5MguFkPOG4bBk/edit?usp=sharing

Demographics/Turnout Data used: https://docs.google.com/spreadsheets/d/1qsWmPyoLF7q30MoULktph9TjKyqDs-dwpC8Yb5gMInI/edit?usp=sharing

[amdcpus]

Template: https://docs.google.com/spreadsheets/d/10semu5xRZ-D6gG7J3d19asXVMnoCvh7G2RNuySOv_RI/edit?usp=sharing

Template with my Utah Poll: https://docs.google.com/spreadsheets/d/1bWqN7yncLyttlivU1hc9FX7xN8jZ4o5MguFkPOG4bBk/edit?usp=sharing

Demographics/Turnout Data used: https://docs.google.com/spreadsheets/d/1qsWmPyoLF7q30MoULktph9TjKyqDs-dwpC8Yb5gMInI/edit?usp=sharing

Thanks!  It looks like you are using the same census data I wanted to use.

No problem. Let me know if you have any questions. I'd also be interested in seeing the raw data too if possible.

[amdcpus]

No problem. Let me know if you have any questions. I'd also be interested in seeing the raw data too if possible.

I'll probably provide a link to the raw data for the survey when it's done.  So far, it has 122 of 500 respondents, so I expect it to be complete Monday or Tuesday - though new respondents have slowed to a crawl in the past few hours.  Some of the cross-tabs look a bit wonky (others look pretty good), but that's to be expected with small sample sizes.

As you know, Google calculates weighted results that are supposed to represent a representative sample of the Internet population based on at least age.  Anyone care to guess the South Dakota results so far - or predict the final results?

Google's weighting is utter crap and for some reason they don't even apply most of the time. Weighted correctly, I expect it to be something like

Trump 42

Clinton 30

Johnson 12

Stein 2

Could you PM the current results? You just have to click "share" and change it to whoever has the link, like this: https://www.google.com/insights/consumersurveys/view?survey=ph4cdzact22bfkpwb34ts46io4&question=1&filter=gen%3AMale%2C%2Cage%3A18-24

[amdcpus]

Google's weighting is utter crap and for some reason they don't even apply most of the time. Weighted correctly, I expect it to be something like

Trump 42

Clinton 30

Johnson 12

Stein 2

Could you PM the current results? You just have to click "share" and change it to whoever has the link, like this: https://www.google.com/insights/consumersurveys/view?survey=ph4cdzact22bfkpwb34ts46io4&question=1&filter=gen%3AMale%2C%2Cage%3A18-24

Google's weights are more like a sample of "adults" or "Internet users" than registered voters.  Best I can tell, they skew young, which helps a certain 3rd party candidate.

I didn't poll Stein because she's not on the ballot.  I polled Castle because he is.  I also didn't include a don't know/undecided or write-in option because, as I understand how GCS works, undecideds can just skip the question, and few people will actually write-in on election day.  I'd rather have no undecided option unless I could follow up and push the undecideds.  That's too expensive.

Anyway, I think I've hijacked this thread enough with technical mumbo jumbo.  I probably should go back to predictions only.

Check your PMs.  Please PM me with any other technical stuff.

Thanks for the info, and yea I meant Castle, not Stein. If you started the poll yesterday, it should be finished around this time tomorrow. Feel free to ask me any questions about the template when you go to weight it.

[amdcpus]

Someone in the Gary Johnson subreddit yesterday reported that they were polled on their cell phone by CNN/ORC. He is in PA so it could either be a PA poll or National poll.

[amdcpus]

Trump +10
Tie
Gregg +6

Almost perfect, Trump +11.

[amdcpus]

Arizona: Trump +6
Georgia: Trump +8
Texas: Trump +12