In a shock poll I saw this morning... Trump is beating all Dems narrowly:
Trump 45%, Clinton 40%. (There is a 20-point Gender Gap; Trump leads by 18 points among seniors.)
* Trump 44%, Sanders 40%. (Trump leads by 10 among independents and by 6 among moderates.)
* Trump 44%, Biden 42%. (Trump leads by 10 among the best educated; Biden leads by 17 among the least educated.)
* Trump 44%, Gore 41%. (Trump leads by 12 among men and by 18 among voters age 50+.)
Why they included Gore I have no idea... but whatever...
Among a subset of registered voters who tell SurveyUSA that they pay "a lot" of attention to politics, the scale tilts to the right: Today it's:
* Trump 54%, Clinton 36%.
* Trump 53%, Sanders 39%.
* Trump 53%, Biden 37%.
* Trump 54%, Gore 36%.
Okay.... it also shows Trump getting 25% of black and 30% of Hispanic vote. Okaaayy... Suuuuurrrree....
But this is where it gets interesting:
Cell-phone and home-phone respondents are included in this survey. SurveyUSA interviewed 1,000 USA adults 09/02/15 and 09/03/15. Of the adults, 900 were registered to vote. Of the registered voters, 58% (522) pay "a lot" of attention to politics. This survey was conducted using blended sample, mixed-mode. Respondents reachable on a home telephone (62% of registered voters) were interviewed on their home (landline) telephone in the recorded voice of a professional announcer. Respondents not reachable on a home telephone (38% of registered voters) were shown a question on their smartphone, tablet or other electronic device. Cell respondents, as is typically the case, vote more Democratic than do landline respondents. Among just the universe of cell-phone respondents, Clinton defeats Trump by 16 points; Sanders defeats Trump by 12 points; Biden defeats Trump by 17 points; and Gore defeats Trump by 17 points. The more cell-phone respondents a pollster includes in its "mix" of voters, the more Democratic the poll results will be.
So among cell phone users Trump gets walloped. We just need to make sure those people vote.
Supposedly SurveyUSA has a good track record with 538 but this is ridiculous.