McClatchy/Marist National Poll: Clinton +1/+2

[Erich Maria Remarque]

Again, Nov 1-3.

[Erich Maria Remarque]

49-34 among Latinos is not likely.

They have Latinos at only 9% of the electorate, too.  I know, subsample MoE, but that's your problem right there.  Trump is only winning white voters by 12 points in this poll, and if that happens, Clinton wins handily.

With the same logic - Hillary is only winning  Latino voters by 15 points in this poll, and if that happens, Trump wins handily.

It cancels out. But 9% latino in demographics is wrong.

34% say they have already voted. Clinton winning 50-42 among people that have already voted, Trump winning 45-44 among those that haven't voted yet.

Is that supposed to make sense?

Just calculated it, it comes out to an overall Clinton +2 lead.

DECIMALS. It might for example be 50.2 vs 43.7

Just a reminder that Obama's average lead in national polls was 0.7% right before election day.

And in 2008 polls were right... You can't cherry-pick single year.

[Erich Maria Remarque]

They have Latinos at only 9% of the electorate, too.  I know, subsample MoE, but that's your problem right there.  Trump is only winning white voters by 12 points in this poll, and if that happens, Clinton wins handily.

Why are you analysing this post selectively? 12 point lead amongst whites is ok but 15 point deficit with Latinos is not ok.

Yes, because 12 points is reasonable - at most he will have a 15-18 point lead with whites, esp with college educated whites swinging to Clinton. However, no one in their right mind thinks that Hillary will underperform Obama with Latino's...and in fact, Latino Decisions puts her margins at "historic" levels, probably +60.

But Whites has much bigger share. 6% of Whites (T+18 as you said) would give extra 70% * 0.06 =5.2%

Latino has about 12%, so 45% (if Clinton +60) would give 12% * 0.45 =5.4%.
And Latino Decision overestimated Obamas margin by 10% points in 2012.

It would cancel out.

[Erich Maria Remarque]

They have Latinos at only 9% of the electorate, too.  I know, subsample MoE, but that's your problem right there.  Trump is only winning white voters by 12 points in this poll, and if that happens, Clinton wins handily.

Why are you analysing this post selectively? 12 point lead amongst whites is ok but 15 point deficit with Latinos is not ok.

My larger point is that given a variety of data, we know that Latino voters will go for Clinton by a much larger margin than 15 points.  

Now it's possible that Clinton does worse with white voters than -12, but if she gets even close to that, she will win because we know approximately what the electorate will look like (and if anything, the early vote indicates that Latino enthusiasm is high this year).  Also, the MoE on the white subsample is a lot lower than the Latino subsample.

But once again MOE of 2% of 72% whites is > than MOE of 15% among 12% Hispanics.

[Erich Maria Remarque]

They have Latinos at only 9% of the electorate, too.  I know, subsample MoE, but that's your problem right there.  Trump is only winning white voters by 12 points in this poll, and if that happens, Clinton wins handily.

Why are you analysing this post selectively? 12 point lead amongst whites is ok but 15 point deficit with Latinos is not ok.

Yes, because 12 points is reasonable - at most he will have a 15-18 point lead with whites, esp with college educated whites swinging to Clinton. However, no one in their right mind thinks that Hillary will underperform Obama with Latino's...and in fact, Latino Decisions puts her margins at "historic" levels, probably +60.

But Whites has much bigger share. 6% of Whites (T+18 as you said) would give extra 70% * 0.06 =5.2%

Latino has about 12%, so 45% (if Clinton +60) would give 12% * 0.45 =5.4%.
And Latino Decision overestimated Obamas margin by 10% points in 2012.

It would cancel out.

They had Obama at 75% in 2012, the exits pegged him at 71. Not bad for gauging a hard to contact group.

This poll has Hillary at 49 with Latino's. That's just not plausible.

I talked about margin. They overestimated Obama's margin by 8-10% (don't remeber correctly).

As I said Trump +12 among Whites is to low. He got the same number, before Reps (and 99% of them was white) coming home.

1% of all Whites = 7% of all Hispanics.

[Erich Maria Remarque]

I'm floored at Trump's strength among Republicans.  Virtually all intra-Party resistance has completely collapsed.  If we've ever seen a time in which a major party has turned into a cult, this is it.

Should be noted that this poll is an outlier though.  No other poll has Trump over 90% among Republicans.  Even when I’m generous to Trump by using the “less smoothing” option at HuffPo (which puts him higher), the trendline has him at 86.8% support from Republicans, as compared to 90.5% support for Clinton for Dems:

GOP

Dems

So yes, Trump is cleaning up among Republicans, but there is still a small band of holdouts within the GOP, which is larger than the holdouts on the Dem. side for Clinton (and larger than the group of Republicans who didn’t vote for Romney in 2012).

EDIT: Caveat to the above: Looks like this poll hasn't been added to HuffPo yes, so I guess that could shift things when they add it.

Not really. It sort of 2012-unkewer misstake, if it is self-reported (I guess) Reps/Dems/Indys.

In this poll Trump is just +3 among Indys. So I guess, more Rep Indys said that they are Reps. Not to talk about self-cancelled MOE in different subgroups etc. (in this case Reps vs Indys among other).

[Erich Maria Remarque]

I don’t understand your point about “more Rep Indys said that they are Reps”.  What do you mean by that?

a) Those who usually say they are Indys (but vote for Trump) said in this particular sample/poll that they are Reps.
b) Those who usually say they are Reps (but don't vote for Trump) said in this particular sample/poll that they are Indys.
c) combination of above

So Trump is doing somewhat higher % among Reps, but somewhat lower % among Indys. It sort of cancels out. This particular poll has R27/I36/D36.

Because as you said, it is just a self-reported feature and often pretty meaningless, besides those cases, when it is way-way off.

[Erich Maria Remarque]

Lol, I only now noted the headline of this poll.
http://www.mcclatchydc.com/news/politics-government/election/article112635048.html

You must be logged in to read this quote.

I say: wow!

Not really, but it is like ~        51% think #LockHerUp

[Erich Maria Remarque]

Alan Abramowitz just mentioned it on twitter. Since their last poll in September, which was Clinton +6, Trump has lost support among whites (+12 from +14), so his numbers improved exclusively due to big gains among non-whites.

Yes, as I said, MOE that cancelled out by each other.  Literally all the polls showed that Pubs/undeciders/Johnson are coming home, but this one shows him losing support among whites and gaining among non-whites? OK, all right!