Welcome, Guest. Please login or register.
Did you miss your activation email?
March 21, 2019, 12:53:38 pm
HomePredMockPollEVCalcAFEWIKIHelpLogin Register
News: Please delete your old personal messages.

+  Atlas Forum
|-+  Election Archive
| |-+  Election Archive
| | |-+  All Archived Boards
| | | |-+  2016 U.S. Presidential General Election Polls (Moderators: AndrewTX, Likely Voter)
| | | | |-+  Google Consumer Surveys 50-state breakdown (search mode)
Pages: [1] Print
Author Topic: Google Consumer Surveys 50-state breakdown  (Read 10921 times)
Seriously?
YaBB God
*****
Posts: 3,036
United States


View Profile
« on: September 10, 2016, 05:18:59 am »

Jesus Christ.

Silver's whole #AllDataPointsMatter nonsense is getting out of control. Adding obvious junk polls under the presumed assumption that an equally polarizing and junkish poll will come along later to balance out the overall result is silliness.
I disagree. He's using them to tack onto a blended sample of data for the state when he runs his projections. He's well aware of the MOE and the tendency for the poll in isolation to be suboptimal, especially with the tiny sample sizes.

The polls are instructive as data points when added to other data points and probably more instructive than Reuters/Ipsos and Washington Post/Survey Monkey as the data is over a week span as opposed to a 1-3 week span (Reuters), or 4 week span (SM).

You could recreate a more representative sample by compounding the surveys into 1, 2, 3 week surveys if you wanted to like Reuters and SM does.

You can likely extrapolate the Reuters/Ispos numbers over a week if you wanted to at this point, but solely if you had access to the underlying data. Reuters doesn't give us this. Google does, down to the individualized weights of the individualized survey respondent.
Logged
Seriously?
YaBB God
*****
Posts: 3,036
United States


View Profile
« Reply #1 on: September 10, 2016, 06:52:14 pm »

no undecideds? pure trash.

I mean, has anyone ever heard of the concept Garbage In, Garbage Out? If you get information from bad and obviously junky sources, guess what? your conclusions are going to be bad and junky.
What makes the sources "junky" to you? Obviously, internet polling is more experimental right now, but it's more likely than not the wave of the future.

These numbers are helpful as data points in poorly polled states, if you put them in aggregate with the other polls of the state. I could see how Silver would use them. If you add the four weeks, you're no different than the Reuters or Survey Monkey polls and have something which, if blended together, can get you a decent poll.
Logged
Seriously?
YaBB God
*****
Posts: 3,036
United States


View Profile
« Reply #2 on: September 11, 2016, 03:23:41 pm »

538 crunched the DC numbers.. They may be close to accurate if they took that leading 2 for Trump and added the 20% to Clinton. (Obviously, the sample sizes are hideous).

AUG. 31-SEP. 6   Google Consumer Surveys    66   LV   
Clinton 48%
Trump 22%
Johnson 9%

AUG. 24-30   Google Consumer Surveys   67   LV   
Clinton 44%
Trump 26%
Johnson 10%

AUG. 17-23   Google Consumer Surveys   
58   LV   
Clinton 54%
Trump 22%
Johnson 8%

AUG. 10-16   Google Consumer Surveys   54   LV   
Clinton 44%
Trump 25%
Johnson 6%
Logged
Seriously?
YaBB God
*****
Posts: 3,036
United States


View Profile
« Reply #3 on: September 28, 2016, 07:54:26 pm »

9/21-9/26 release

Img
« Last Edit: September 28, 2016, 08:09:59 pm by Seriously? »Logged
Seriously?
YaBB God
*****
Posts: 3,036
United States


View Profile
« Reply #4 on: September 28, 2016, 08:03:05 pm »

Idaho!?
That one is wrong. Have to check the spreadsheet and recast.
Logged
Seriously?
YaBB God
*****
Posts: 3,036
United States


View Profile
« Reply #5 on: September 28, 2016, 08:10:44 pm »

Idaho was the only error. I had the wrong postal code to crunch the numbers, it was duplicating Iowa's numbers.
Logged
Seriously?
YaBB God
*****
Posts: 3,036
United States


View Profile
« Reply #6 on: October 05, 2016, 11:08:04 pm »

9/27-10/3 breakdown with margins sorted by state

Img
Logged
Seriously?
YaBB God
*****
Posts: 3,036
United States


View Profile
« Reply #7 on: October 06, 2016, 09:44:56 pm »

46% Hillary Clinton (D)
21% Donald Trump (R)
9% Gary Johnson (L)

Just saw this on twitter... B-b-but I thought NH was supposed to be more competitive than VT and MA this year! Sad
They asked like 74 people on line at the local Wal*Mart who they'd support for President. Some of htese polls have such a small sample size to be practically meaningless.
Logged
Seriously?
YaBB God
*****
Posts: 3,036
United States


View Profile
« Reply #8 on: October 11, 2016, 12:06:34 am »

I honestly think that GCS has a systemic problem polling the South.
Or DC or the midwest.

These polls are basically 4-question polls and look at geolocation to determine where the voter is going to vote, which is why for a while, they got terrible results in DC.

Obviously 2 of the 50 polls should fail as a matter of science at 95% confidence and some of the MOEs are gigantic (especially in the smaller states), which make these polls dicey.
Logged
Seriously?
YaBB God
*****
Posts: 3,036
United States


View Profile
« Reply #9 on: October 12, 2016, 01:54:01 am »

Img
Logged
Pages: [1] Print 
Jump to:  


Login with username, password and session length
Logout

Terms of Service

Powered by SMF 1.1.21 | SMF © 2015, Simple Machines