Pollster.com

Articles and Analysis

 

POLL: SurveyUSA Ohio


SurveyUSA

Ohio
Clinton 50, Obama 44... McCain 59, Huckabee 28, Paul 8

 

Comments
pf:

that's Clinton 50, Obama 44

____________________

Mark Silk:

I believe that should be Clinton 50, not Clinton 40.

____________________

David:

Uh, I think that's supposed to be Clinton 50, Obama 44

____________________

sote:

Nice try Pollster, drinking the Obama juice as well....

____________________

Bob Evans:

OMG! Billary is down to 40! HAHAHAHAHAHAHA!

Nice firewall. That's too funny. Billary supporters are going to cry a river tonight! Time to get blitzed in Youngstown!

____________________

infogiant:
____________________

daniel:

When is pollster going to update the national poll? Its been nearly a week

____________________

C.S.Strowbridge:

"Nice try Pollster, drinking the Obama juice as well...."

Yeah. They did it on purpose and it wasn't just an honest typing error.

____________________

polsaa:

This survey has 23% of African Americans voting for Clinton. That's not going to happen. Instead of 70-23 it will be more like 85-15 or more for Obama. Otherwise this looks pretty reasonable.

____________________

DJShay:

"This survey has 23% of African Americans voting for Clinton. That's not going to happen."

I don't know. Stephanie Tubbs Jones has been campaigning pretty hard for Clinton.

____________________

Anonymous:

um it says 50 percent clinton... dont tell me the polling websites are againist her also... lol... this media is sooo funny!!

____________________

Freedom Fries:

Even if it is Billary 50, she her lead is eroding . Plus, O'bambi will crush her in TX

____________________

Greggie:

She hasn't gotten 23% of the AA vote anywhere in the country. John Lewis campaigned hard for her in Georgia, too. So did Maxine Waters in California. Etc. Obama has outperformed his polls among the AA community in every state that has an AA community so far, I think (I may be missing an outlier somewhere). But it's a pretty convincing trend...

____________________

RS:

The "Clinton 40%" is a TYPO.

Look at the Ohio polls page on this site, it has Clinton 50, Obama 44.

Jeez, give the good folks at Pollster a break, will ya?

____________________

craigc:

You posted this incorrectly...It is clinton 50 and Obama 44

____________________

Huckabee continues to rise in each state as the election approaches. When people read his positions against IRS and Federal Reserve - both privately owned - they get excited about America. Some become proud of America for the first time.*

*Attributed to Mrs. Obama (don't want to anger Hillary.)

____________________

Freedom Fries:

Even if it is Billary 50, she her lead is eroding . Plus, O'bambi will crush her in TX

____________________

Robert Avila:

Why did Pollster go back and alter the graph for Huckabee in the Texas polls? Yesterday, and all week, they had a straight line passing ONE low, heavy outlier. Today, they show a sharp spike downward, then a sharp spike back up from there. This is inconsistent with other graphs, all of which are smoothline. Uh, is pollster also anti-Huckabee? Perhaps they are bought and owned by the same people who own the Federal Reserve?

____________________

Engaged College Student:

*raises hand.

How are the pollsters accounting for the "cell phone only" youth vote in Ohio?

Are they simply ignoring it?

Thanks.

____________________

Phil:

Clinton is down to a 6% lead in Ohio?

Ouch. How much has that closed in the last week?

____________________

Bob Evans:

The fat lady is singing for Billary! Time to add another to the Clinton Body Count! Looks like Maggie Williams will suddenly get "heart problems" or a wicked case of "pneumonia." The sweet stench of failure awaits all the suckers who bought into Billary! I spent $2300 and all I got was PMS rage and boo-hooing!

HAHAHAHAHA

GOBAMA!!!!

____________________

As MANY of you pointed out, we had a typo in the initial post of this. Sorry for that-- we try never to make mistakes but sometimes they do slip through. Thanks to those who immediately pointed it out so we could fix it. (It is now fixed both here and on the front page. If it doesn't look fixed either place, try refreshing the page or possibly clearing your cache.)

Charles

____________________

illinoisindie:

Usually I drink the SUSA Kool-aid but some of these numbers are adverse to demographic voting trends like the HRC&BO basically being split in the independants and a 75/25 african American vote. This one might need a grain of salt. Anyone know how Ramussen is splitting the demographic

____________________

Robert Avila asks about a change in the Huckabee graph for Texas and in the process charges we are anti-Huckabee.

The change in appearance is due to the addition of new polls. ALL our charts dynamically update when new data come in. With fewer recent polls, the trend estimator draws a best fit between the late 2007 data and the cluster of recent 2008 polls, in the process largely ignoring the one 20% poll from 1/30-31/08. Perhaps paradoxically, since MORE recent polls became available there is now enough data to fit a better trend for the 2008 data. That fit to recent data then allows the trend to bend more in response to the 1/31 outlier. That peculiar bend only occurs because the data are very sparse in January, so there are few points to control the trend in that period. So the "kink" in the trend line is due to the combination of an outlier when few other polls are nearby and the plentiful data in recent days to fit the trend in the second half of February.

But I fail to see the logic that makes our analysis anti-Huckabee based on one transitory kink in the trend when the recent trend is clearly up in February and where the trend obviously fits the recent data.

And if we were "owned by the same people who own the Federal Reserve" I can assure you Mark and I would be enjoying a tropical island rather than DC and Madison in February!

Charles

____________________

BitJam:

Charles,

Is the algorithm you use for the trend lines posted anywhere? I've looked around but I've been unable to find it.

____________________

Bob Evans:

Does anyone know when the next SUSA polls come out that will show Obama crushing Billary by 10 points in both states? After tonight's debate, it's so over for Billary. Slick Willy is going to have to get a real job now!

GOBAMA!!!!!!

____________________

Mark Lindeman:

Charles, I had assumed that the Federal Reserve business was satirical -- but come to think of it, how do we know that you aren't enjoying a tropical island? In cyberspace, no one can see you sunbathe....

The trend estimates right now seem to be slightly understating Obama's progress in Ohio, but I'm not sure that's a weakness. Maybe it is. If five polls come out on the same day, do they have relatively less immediate impact on the trend line than if they came out over three days? One could argue either way what the best behavior would be. (Clearly I don't spend much time wondering about trend estimates, or else I would already know the answer to this question.)

____________________

New Pennsylvania poll from Quinnipiac has Clinton up by 6, down from 14 on Quinnipiac's last poll. It's not looking like much of a firewall for her anymore.

____________________

David :

I predict 60-40 Texas, 55-45 Ohio, 65-35 Vermont, 55-45 Rhode Island.....

all for your new and still undefeated Champiiiiooonn of the worrrrlldd......

Barack OooooooBBBaaaaaammaaaaaahhh
Obama!

____________________

Dave S:

Dear lord, could people cool it with the paranoia? It never ceases to amaze me that more and more, people on the internet seem to assume that any information which doesn't jibe with them, or any error or omission, no matter how incidental, must be the product of some vast conspiracy.

____________________

Brian:

The election is over and Obama will be the Democratic nominee for President.

Good luck John Mccain because you're in for a major league beating!!

These red states/lean red states in presidential elections are now in play

Alaska (sleeper pick)
Nevada
New Mexico
Colorado
Mizzou
Virgnia
Iowa

____________________

BitJam-- I've described it in several places but we should have a FAQ available on this. I'll see if I can't get one put together and posted. A brief answer is that the trend is a local regression fit, which allows the trend to follow the data but with some constraint on how "rough" or "smooth" the estimate is. I often fit a "more sensitive" trend to compare to our standard fit, in order to see if the amount of smoothing matters. The way the local regression is estimated gives great weight to points close by and little or no weight to points far away, and reduces the weight given to outliers (though outliers near the end of the data can pull the trend more than you might like.) In some cases I estimate the trend after taking out "house effects" due to different pollsters. That hasn't been possible with the primary polling because many pollsters only poll one state, and therefore we don't have enough data on each polling organization to estimate an effect for each one. This is an area of ongoing work for us. I hope that helps answer your question.

Charles

____________________

Mark Lindeman--- Believe me if I was doing this on the beach I wouldn't be wearing this big parka here at 9 degrees this morning!

Mark also asks "If five polls come out on the same day, do they have relatively less immediate impact on the trend line than if they came out over three days? One could argue either way what the best behavior would be."

That's a good question. if they all come out on one day, then the trend estimator will, all things being equal, try to go through the mean of the five polls, necessarily going over some and under others. If they were spread out a bit in time, then they would influence the trend IF they also were consistently rising or falling with time. But if they were randomly scattered with respect to time, the trend would properly be flat over those 5 polls. Of course all isn't equal: there are previous polls that have influenced the trend estimate before the new 5 polls appeared, and the earlier polls will continue to have some influence (especially if they are near in time and numerous). Given the past trend the five polls on the same day will tend to produce a trend that goes through their middle from the previous data points. Given a random spread of the same 5 polls over a few days, the trend will try to rise (or fall) to the beginning of the five and then go flat through them since the 5 new ones have no trend at all. Of course, "only 5 polls" still won't completely dominate the estimator given a good number of previous polls, so the effects here will be less dramatic than I describe.

Charles

____________________

BitJam:

Thanks for the info Charles.

I've worked on various curve fitting and data smoothing algorithms over the years. I'm looking forward to seeing the nitty gritty details of what you are using.

Have you ever considered using a piece-wise parabolic (or piece-wise cubic) fit instead of piece-wise linear? There are simple alogorithms available that don't drastically increase the number of free variables. For example, if you convolve your piece-wise linear fit with a tophat function then you end up with a smoother, piece-wise parabolic fit. The one free variable you add is the width of the tophat but that can easily be tied to the distance between the knots.

____________________

Sami:

Obama is far ahead of Clinton in Ohio. The rallys show three times the crowd for Obama in Ohio and Texas. In addition, Obama is outraising Clinton FOUR TO ONE! This really is the true poll. The only reason that Clinton could catch up to Obama would be to manipulate the vote and to not count votes intentionally as has happened in previous elections. The third reason would be debolt machine tampering.

____________________

Sami:

Obama is far ahead of Clinton in Ohio. The rallys show three times the crowd for Obama in Ohio and Texas. In addition, Obama is outraising Clinton FOUR TO ONE! This really is the true poll. The only reason that Clinton could catch up to Obama would be to manipulate the vote and to not count votes intentionally as has happened in previous elections. The third reason would be debolt machine tampering.

____________________

Sami:


Obama is far surpassing Hillary in Ohio and in Texas. Obama draws five times the crowds and money! The only ways Hillary has a chance is vote manipulation, not counting votes and debolt machine tampering.

____________________

Sami:

Obama is far surpassing Hillary in Ohio and in Texas. Obama draws five times the crowds and money! The only ways Hillary has a chance is vote manipulation, not counting votes and debolt machine tampering.

____________________



Post a comment




Please be patient while your comment posts - sometimes it takes a minute or two. To check your comment, please wait 60 seconds and click your browser's refresh button. Note that comments with three or more hyperlinks will be held for approval.

MAP - US, AL, AK, AZ, AR, CA, CO, CT, DE, FL, GA, HI, ID, IL, IN, IA, KS, KY, LA, ME, MD, MA, MI, MN, MS, MO, MT, NE, NV, NH, NJ, NM, NY, NC, ND, OH, OK, OR, PA, RI, SC, SD, TN, TX, UT, VT, VA, WA, WV, WI, WY, PR