Another Iowa Poll Has Hillary And Romney Leading, Huckabee Surging

Comments (22)

elrapierwit wrote on October 18, 2007 11:22 AM:

Would you please provide the MOE for this poll?

Any time you have a headline asserting that any candidate is 'leading' in a poll it should be mandatory to provide the MOE...that is the only thing that validates the asserts you continue to make with your headlines... which have been shown many times to be questionable.

david knowles wrote on October 18, 2007 11:26 AM:

Brownback is out. Anyone doubt that this is due to the Colbert Effect?

elrapierwit wrote on October 18, 2007 11:26 AM:

Would you please provide the MOE for this poll?

Any time you have a headline asserting that any candidate is 'leading' in a poll it should be mandatory to provide the MOE...that is the only thing that validates the asserts you continue to make with your headlines... which have been shown many times to be questionable.

OIC....once again Kleefeld your headline is FALSE!!

The margin of error is 4...which means Hillary is not leading anything. At best she is tied with Obama for the lead and possibly trailing him by 3 points.

Why don't you just report the facts and stop with all these ERRONEOUS headline, don't you have any pride when it comes to credibility?

As it stands now, whenever your byline is underneath any headline proclaiming a lead for Hillary it is immeadiately suspect, that is the type of track record you have earned.

bob wrote on October 18, 2007 11:42 AM:

A Huckabee surge is good and bad news. I think he's probably better than Romney and Giuliani in some ways, as he's not either obviously insane or a total phony who would say or do anything to be get elected and then re-elected. I have a feeling the odds of war with Iran are lower (although I'm sure he's said similar things to the other two).

The bad news is that he may well be more electable than the other two. Giuliani eventually will come off as unhinged I think, if he somehow manages to win the GOP nomination without the Christian right (a dubious proposition), and Romney has to overcome both his hypocrisy and his faith.

Anonymous wrote on October 18, 2007 11:42 AM:

The margin of sampling error is ±4 percentage points.

http://strategicvision.biz/political/iowa_poll_101807.htm

hadenough wrote on October 18, 2007 11:52 AM:

"elrapierwit wrote on October 18, 2007 11:26 AM:
Would you please provide the MOE for this poll?
....
The margin of error is 4...which means Hillary is not leading anything. At best she is tied with Obama for the lead and possibly trailing him by 3 points."

Would someone please tell obamavangelists what margin of error is?

At best Hillary is not only leading but crushing obama. With the MOE Hillary could be at 32% and obama could be at 19%.

And yes I know the actual pollsters who poll for a living have no clue what they are doing. Only obamavangelists do. Actual pollsters who poll for a living don't look at their own crosstabs and don't look at margin of error the poll that they conducted and figure all that in to the final numbers they come up with. Only obamavangelists do.

Jane wrote on October 18, 2007 12:01 PM:

Elrapierwit:

You misunderstand and misuse the margin of error. While with these numbers and a margin of error of 4 it may be the case that the true numbers (if everyone was counted rather than simply a sample) could be Hillary 24 and Obama 27 , it is also possible AND EQUALLY LIKELY that the true numbers are Hillary 32 and Obama 19.

This simplifies but makes the essential point that the margin of error swings both ways.

pacc wrote on October 18, 2007 12:06 PM:

So refreshing to read hadenough and Jane's counterpoint to the crazy paranoia that always seems to creep into O-Bomb-A cheerleaders' posts.

elrapierwit wrote on October 18, 2007 12:11 PM:

Thanks Jane,

While I did use the MOE to swing to the 24/27...I deliberately did not swing the other way..simply to be as biased as the headline.
Kleefeld's headlines are consistently misleading and aggravatingly so.

DTM wrote on October 18, 2007 12:11 PM:

I'm not sure what it means to claim that crosstabs and MOEs are "figured" into the "final numbers". Those items are just more information about the actual results of the polling process. Specifically with respect to MOEs, the point of course is that the "final numbers" are more properly conceived of as a range, rather than a particular number (and even then, a certain percentage of polls will even get the range wrong).

Of course, in general the common problems within the public discourse about polls is usually not caused by the pollsters (at least not by neutral pollsters). Those problems are usually caused by journalists, campaigns, or other people commenting on the polls (when, for example, they fail to report MOEs, imply that polls are strongly predictive, ignore the likelihood of random fluctuations, and so on).

Keith wrote on October 18, 2007 12:13 PM:

Without more information on the internals of this poll, it's irresponsible to speculat about what it means. It shows HRC leading by 5 points. Just outside the MOE. Without more, the most you can say is that this poll confirms that Iowa is a TIGHT race.

pacc wrote on October 18, 2007 12:18 PM:

More for the O-Bomb-A the phony file...

Senator Now Seeks Lobbyists By Tory Newmyer
Roll Call Staff
October 18, 2007

Even as Sen. Barack Obama (D-Ill.) has intensified his attacks on Sen. Hillary Rodham Clinton (D-N.Y.) for her ties to K Street, he has been reaching out to lobbyists to provide volunteer manpower in early primary states.

http://www.rollcall.com/issues/53_45/news/20538-1.html

elrapierwit wrote on October 18, 2007 12:33 PM:

So let's see now we are suppose to believe that volunteer manpower is the same as corporate dollars...millions and millions of dollars.

Hillary's team is always trying to blur the line.

The truth is that Hillary has more lobbyist money than ALL of the ENTIRE presidential field combined..i.e. Dems AND repubs.

Volunteer manpower simply does not major up to that.

Marge Inovera wrote on October 18, 2007 12:42 PM:

Nice explanation of the margin of error when looking at the difference between the support for two candidates here.

The question is what is the margin of error on the difference between the values for Senator Clinton and Senator Obama. The formula on the above page shows that the error on the difference is 3 percentage points. Thus, the difference is 5% +/- 3%. This means that there is a 95% chance that Senator Clinton is leading.

This assumes that the systematic error (e.g. bad formula for who is a "likely" voter) is small compared to the sampling error.

Keith wrote on October 18, 2007 1:28 PM:

Marge:

It seems to me this analysis goes both ways. There is an error that the sampling error of support for HRC is overstated by the MOE and that support for Obama is understated by the MOE. So I'm not sure reducing the MOE +/- is really saying much in this case (As I said above, all we can say is the race is TIGHT at this point).

And, again, we have no idea on how they selected these individuals, or the mix (weight) of individuals, etc. Without that information, it's impossible to draw any firm conclusions about this poll (other than the race in Iowa is tight). At least in my opinion.

DTM wrote on October 18, 2007 1:34 PM:

Marge,

The suggestion on that page is: "When reading political polls, remember that the margin of error in an estimate of the 'gap' between the two leading candidates is roughly twice as large as the poll's reported margin of error ...." That, of course, would make the proposed MOE on the difference 8% (twice the reported MOE of 4%).

I am not saying you are wrong, however. I just think you need to elaborate on how you got your result (I suspect it may have something to do with both Clinton and Obama being far from 50%, but I will let you provide the explanation).

kjoe wrote on October 18, 2007 1:48 PM:

I thank Jane and Hadenough for their posts. Their math is right.

I am a big fan of Obama---I thought he made a good point on Leno last night about Hillary being careful about declaring "mission accomplished". It was a good appearance.

Let's keep Jane's and Hadenough's math in mind during the coming weeks. For now, I hate to say---Hillary's 11 point lead in yesterday's Rasmussen poll, which, with a margin of error could be even larger will be the one most widely quoted.

There will continue to be polls, and polls, and then some more polls.

And then, people will start voting, and there will be more polls about what they did, and what others will do.

Marge Inovera wrote on October 18, 2007 1:58 PM:

"There is an error that the sampling error of support for HRC is overstated by the MOE and that support for Obama is understated by the MOE."

Statistical error is more complicated than that.

When you tell a competent machinist to cut something to 1/2 an inch plus or minus 1/16 of an inch, the piece that you get will be between 3/16 and 5/16. The tolerance 1/16" is an absolute limit.

Statistical error is different. The usual value used indicates that the correct value is 68% likely to be within the error of the reported value and 95% likely to be within 2 times the error. There is a small chance that the observed result is vastly different from the correct result. This error scales like the inverse of the square root of the number of people polled. This error is also the standard deviation of a hypothetical infinite set of repetitions of the poll.

One difference between statistical error and tolerance (in the machinist's sense) is how they act when you calculate other values from them. If you stack two pieces of metal that are 1/2" +/- 1/16", you get something that is 1" +/- 1/8". But if you add two (independent) numbers with statistical error like 10%+/-4% and 15%+/-4% you get 25%+/-5.7%. The errors do not just add up.

You are right about the importance of how they select individuals.
The only way to address the systematic errors is to look at several different polls. We hope, possibly in vain, that their methodologies for selection are sufficiently different that systematic errors tend to average out. If the pollsters all make the same systematic errors, then the polls give us the wrong answer even in aggregate.

elrapierwit wrote on October 18, 2007 2:36 PM:

Clinton leads in polls but had yet to best Obama in the money race. She did so in the third quarter, but not nearly as decisively as the campaign had hoped.

In the end, Clinton reported raising $28 million between July 1 and Sept. 30 and having $50 million in cash-on-hand.

Of that, about $34 million can be spent on the primary, while $16 million is for the general election. Clinton also reported carrying $2 million in debts, driving her ready cash figure down to around $32 million.

Obama’s report, the last to go public, showed him raising $21 million during the third quarter and ending it with $36 million in cash.

pacc wrote on October 18, 2007 2:42 PM:

elrapierwit wrote:

So let's see now we are suppose to believe that volunteer manpower is the same as corporate dollars...

The point, O-Bomb-Lette, is not that the volunteer lobbyist's in-kind value is or is not equal to direct cash payments, but rather that O-Bomb-A continues to run a rather hypocritical and quite shady campaign.

Jan wrote on October 18, 2007 3:08 PM:

For those of you demanding that Eric give you the inside numbers, why don't you just go to the polls yourselves?

I love stats so I love polls.
Having said that, polls are exactly like Las Vegas odds (which I also like because I'm a sports fanatic).

Polls (and odds) DO NOT tell you who will win the game. But professionals DO put numbers together to help betters see what their chances are of winning.

Right now, Clinton supporters like the chances. Right now, others don't.

For me, the only important part of the polls right now are the trends.
Take the same poll and see what that poll said in the past. It's these trends right now that can currently offer some measurable info.

Since I'm liking what I see in the polls, I'll shut up.

But to question respectable polling because you don't like the numbers is like ignoring the numbers that Vegas casinoes come up with when you place your bet, because they didn't give your teams the right odds to win to the Superbowl.

Absolutely nothing wrong will still betting on your team, but that doesn't mean the odds are off.
(And, again, nor does it mean that the team with the best odds WILL win the game, just to be fair.)
It's your perogative not to bet on the team with the best odds, but that doesn't make the numbers less valuable.

For those who want the inside numbers, they are easily available to you.
Try pollingreport.com for a start.

DTM wrote on October 18, 2007 4:11 PM:

Jan,

But horse-race polls are not in fact the same as odds. Odds are a statement about the likelihood of something happening in the future. Horse-race polls are very different: they report information about something happening right now, namely the current preferences of the people polled.

Of course, one can find many people who try to derive odds with respect to the actual election from such polls. But that requires all sorts of inferences: the poll itself is not purporting to state odds. And obviously, the farther away the actual election, the less predictive the poll, because the more and more likely it is that people will arrive at different preferences by the time the actual election arrives.

And all that is ignoring the likely voter problem ... but hopefully the distinction between odds and polls is already very clear.

Post a comment

Name:
Email Address:
URL:
	Remember personal info?
Comments: (you may use HTML tags for style)