Talk Elections

I conducted this poll with Google Consumer Surveys with 500 registered Utah voters. I weighted the data using Utah's age and gender demographics (I broke the data down into 12 groups, Google Consumer Surveys gives 6 age groups and I used them for each gender and weighted each accordingly using Utah's demographics). I additionally weighted by age group for percent of the electorate and turnout by age. This is why the numbers are changed from last time I posted, and IMO, are much more accurate this time.

Results:

Trump 29%

McMullin 23%

Clinton 21%

Johnson 7%

These numbers are pretty in line with Monmouth's and Y2 Analytics recent Utah polls.

Poll was conducted October 12-13 with 500 Utah voters.

Excel sheet of how the data was calculated can be downloaded here:

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

They're polls have been trash. I'll take this one with a grain of salt

Wow, that is fascinating. How much does a poll cost, and where can I find a place to buy them?

According to Google Consumer Surveys (https://support.google.com/consumersurveys/answer/2447244?hl=en&ref_topic=6194684), a 1 question survey with age, gender and location targeting should cost 15 cents per respondent.  So a 500 respondent sample should cost $75.  A geographically targeted 2-10 question survey, necessary for a LV screen and additional demographic info, costs $1.50 per respondent.

You can create your own survey by clicking on the Create a Survey tab here (https://www.google.com/insights/consumersurveys/how).

Swap McMullin and HRC, and it becomes much better.

What the heck?  I'm trying my luck at a South Dakota poll, since many here want to see one.  According to ballotpedia, Stein isn't on the ballot, but Johnson and Castle are, so assuming Google accepts my survey, the choices will be Trump, Clinton, Johnson and Castle (randomized), plus a I am not registered in South Dakota option to try to weed out non-registered voters and out-of-state residents.  I didn't want to add an undecided/wouldn't vote option, because that would probably lead to really crappy results with 30% undecided or something.  Building a multi-question survey with a likely voter screen would be too expensive.

I can't weigh age against the 2012 exit poll because there was none.  Should I just use census data?  (What did you use, amdcpus?).

I blame you, Likely Voter: I posted this on Sunday, and you deleted the thread!

Huh? This is the first time I released the numbers.

     McMullmentum!

A really neat project, and you'll always find people hungry for more numbers here. Thanks for sharing!

word!

cyinc will probably release a SD poll within the next few days and I know two other people doing a Montana and Utah poll, as well as two more who might do a Georgia and Pennsylvania poll.

Really interesting, thanks for sharing.