CNN - Trump and Sanders lead IA

[Statilius the Epicurean]

President Trump.

[cinyc]

Since November/December, we had the Cruz surge where he led in Iowa. Now, Trump is. So, while his lead has decreased since then, it's increased drastically in the past few days.

That depends on which Iowa poll released today you believe.  Some show the race still close.  Others show Trump with a double-digit lead.  It's too soon to tell, really, though at the surface, Cruz has had a bad week so far and Trump has generally had a good week (other than the 2 Corinthians thing).

[Mike Thick]

Since November/December, we had the Cruz surge where he led in Iowa. Now, Trump is. So, while his lead has decreased since then, it's increased drastically in the past few days.

That depends on which Iowa poll released today you believe.  Some show the race still close.  Others show Trump with a double-digit lead.  It's too soon to tell, really, though at the surface, Cruz has had a bad week so far and Trump has generally had a good week (other than the 2 Corinthians thing).

True, and

that Cruz is leading. However, given the events that you mentioned I'm inclined to believe the pro-Trump results.

[Attorney General, LGC Speaker, and Former PPT Dwarven Dragon]

IA: The only poll that matters is one on election day.

[Landslide Lyndon]

Sorry but it seems to me that CNN polls' only goal is ratings and clicks.
If they showed what every other poll shows then nobody would care.
But by showing completely unexpected numbers they have every political junkie talking about them.

[Holmes]

For what it's worth, this poll is predicting huge (yuge) turnout, muuuch higher than 2008 and 2012. And has the Democratic caucus about 50/50 male/female. Those factors favor Trump and Sanders.

[The Other Castro]

Men: Sanders leads 63-32-4
Women: Clinton leads 56-38-4
Whites: Sanders leads 49-45-3
Democrats: Sanders leads 48-47-3

[Bakersfield Uber Alles]

So Iowa may be anywhere from +10 Clinton to +10 Sanders. Good to know.

I don't always agree with you, but here I do. No idea who's winning in Iowa.

It's a generic answer but it'll really depend on who brings out their supporters. Clinton is crushing it among those who are regular caucus goers, and Sanders is among those who aren't, so he definitely has the steeper hill to climb.

Definitely. Obama got first timers to turn out, but we'll see about Sanders in 11 days.

Depending on how close it is, the O'Malley voters could swing things by a couple percent which clouds things even more.

[Eraserhead]

I'm inclined to give more weight to CNN/ORC than I do a couple of garbage college polls.

But yeah, we'll see. It's basically all about turnout at this point.

[Mehmentum]

So Iowa may be anywhere from +10 Clinton to +10 Sanders. Good to know.

I'm splitting the difference and calling it a tossup.

[Zanas]

I'm surprised no one pointed that before :

On the Pub side, the sample is 266 likely caucus goers. That is a way smaller sample that any poll in 2015, except from two... Gravis polls. It makes for a 6+ pt margin of error, so a bit of caution should be needed here, but hey I don't need to remind y'all psephologists that...

On the Dem side, nothing really better. A sample of 280 likely caucus goers. Again, the only smaller samples have been those two Gravis polls and a Loras one a year back. Again, bit of caution ?

I'm not saying Trump and Sanders are not leading, or have not gained or anything, but the sample size should definitely make this poll count way less than other recent polls in any reasonable aggregator.

[Likely Voter]

Polling a caucus is hard and it is all about turnout and organization. If new voters show up in yoooge numbers then Trump and Sanders will win. This poll shows what that scenario looks like.

[Zanas]

Again with the above reservations, Sanders leads Clinton 51-45 on who's the best to handle healthcare. Seems like the fallacious attacks of the Clinton camp on how Sanders would simply repeal Obamacare and leave all Americans without healthcare failed or even backfired.

[Starbucks Union Thug HokeyPuck]

Men: Sanders leads 63-32-4
Women: Clinton leads 56-38-4
Whites: Sanders leads 49-45-3
Democrats: Sanders leads 48-47-3

That can literally not be true.  Bernie is not winning Iowa by more than he is winning whitey.

[cinyc]

Polling a caucus is hard and it is all about turnout and organization. If new voters show up in yoooge numbers then Trump and Sanders will win. This poll shows what that scenario looks like.

FWIW, the sample size of the likely caucus goers in this poll is really small - 266 on the Republican side and 280 on the Democratic side.  That makes the declared MoE +/-6 in each contest.  The Emerson poll (showing Clinton with a 9 point lead) has similar sample sizes.  Loras and Monmouth/KBUR have larger sample sizes.

[Lief 🗽]

Sorry folks, but the turnout models on these two polls are really bad. Not sure why CNN doesn't hire some decent polling firms.

[The Other Castro]

Men: Sanders leads 63-32-4
Women: Clinton leads 56-38-4
Whites: Sanders leads 49-45-3
Democrats: Sanders leads 48-47-3

That can literally not be true.  Bernie is not winning Iowa by more than he is winning whitey.

The total and the white numbers both have +- 6% margins of error so who knows. Maybe his nonwhite numbers are very strong, but CNN didn't release those.

[psychprofessor]

Sorry folks, but the turnout models on these two polls are really bad. Not sure why CNN doesn't hire some decent polling firms.

Considering the 2008 Iowa caucus was 57 / 43 women / men, and this is 50 /50, yes, this is not really representative.

[Amenhotep Bakari-Sellers]

Clinton has the same problem with white males as she does in GE matchups and Iowa is no exception. If Dems do get an unscandalised nominee, it is only a positive.

[psychprofessor]

Clinton has the same problem with white males as she does in GE matchups and Iowa is no exception. If Dems do get an unscandalised nominee, it is only a positive.

Eh, she may have a problem with white males but in a general election matchup she would really have a problem with white, non-college educated men. But they are a decreasing share of the overall general electorate. The thinking is she would make up ground with white women, Hispanics and hold the line with Blacks. I think if Trump is the nominee you see a huge swing of non-white voters (Hispanics and Asians and keep Blacks at Obama levels) for Clinton.

[Ebsy]

Yeah, this poll isn't very good.

[Zanas]

Sorry folks, but the turnout models on these two polls are really bad. Not sure why CNN doesn't hire some decent polling firms.

Considering the 2008 Iowa caucus was 57 / 43 women / men, and this is 50 /50, yes, this is not really representative.

This is not 50/50. They give an 8 pt MoE for the Dem female sample, which points to roughly 150 women, and an 8.5 pt MoE for the male sample, which points to roughly 130 men. This is 53.5/46.5, which is still probably too masculine.

The whole subsample section of this poll is a laugh. It's 50 pages well worth reading for anyone with any notion of statistics !

[Skye]

What a huge outlier.

[Starbucks Union Thug HokeyPuck]

Math has your answers, folks!

Men: 63-32 Bernie
Women: 56-38 Hillary

Actual (hypo, but the sampling was equal) - 50/50 men-women, 140 EACH

UNSKEWED TO 2008 REALITY 120 MEN, 160 WOMEN

Among men: 76 vote Bernie, 39 vote Hillary
Among women: 61 vote Bernie, 90 vote Hillary

Overall: Bernie 137, Hillary 129

49 - 46

So there ya go.  Unskewed or whatever.  It's a dead heat by 2008 turnout.  

[Ebsy]

I don't see the point in "unskewing" polls. If the methodology is off in one way, it is more than likely off in others.