New Poll: Trump and Clinton deadlocked in North Carolina

[WI_DEM]

Washington (CNN) — Donald Trump and Hillary Clinton are deadlocked in North Carolina, according to a new poll released Tuesday.

The poll from Elon University found Trump leading Clinton by 1 point among likely voters, 44% to 43%, a statistical tie, within the 3.86-point margin of error. Libertarian Gary Johnson drew 6% support, while another 6% said they were undecided.

Clinton and Trump maintain large advantages in North Carolina with key constituencies -- Clinton receives the support of 98% of black voters, while Trump is backed by 65% of white voters. The gender divide is less severe, however, as Clinton leads Trump among women by 6 points while Trump's advantage among men is 10 points.

The Elon University poll of North Carolina surveyed 644 likely voters, with a margin of error of +/- 3.86 points. It was conducted from September 12-16.

http://www.cnn.com/2016/09/20/politics/north-carolina-poll-hillary-clinton-donald-trump/

[WI_DEM]

1. BTW, PPP will have a new NC poll out tomorrow.

[Charles Bukowski]

2. This poll caught HRC at the lowest point in her campaign

while missing Mein Trumpf's latest meltdown.

All things considered, a very good result.

[LenaBaby61]

5. This poll caught HRC at the lowest point in her campaign.

And when Mrs. Greenspan, amongst others, were heavily criticizing her for not telling THEM that she had Walking Pneumonia and wasn't near death.

[triron]

3. Internals

said less than 10% under 30, And less than 25% under 40. Don't think that reflects turnout.

[LenaBaby61]

4. New Poll: Trump and Clinton deadlocked in North Carolina.

"Internals said less than 10% under 30, And less than 25% under 40. Don't think that reflects turnout."

I noticed that also when I checked the polls internals.

[triron]

6. HRC

won the under 40s by 64-36. So if turnout is boosted in that age group by about 10-15% that should give her a slight advantage. (even by 5% would make it even)

[triron]

7. Looked more

Last edited Tue Sep 20, 2016, 06:34 PM - Edit history (1)

closely at internals. I noticed 12 out of 644 LV (I think) did not answer the question about how old they were.
So best thing is to assume they are distributed fairly uniformly among voter sample.
Using that assumption and using percentages in each age group and how they said they would vote I could not replicate the result in this poll. Clinton performed better (about 44,4% rather than 42.8). If the turnout model is estimating too low on the under 40 age group where she wins 64% of the vote (for example pretend turnout is 33% of total voter turnout instead of only 24.3% (which this poll uses) she rises to 45.8. In both scenarios she beats Trump barely in the 1st) according to my calculations (but clearly in the latter --about +3). Since the absolute numbers (at the beginning of the write up) do not lie, my guess is that these percentages in the internals have the effect of rounding errors.