More oh-eight: A new SC poll

Jon — Thu, 26 Jul 2007 13:18:08 GMT

This poll's methodology is terrible (as a former polling analyst, I know).

I don't know who's winning in SC, but this poll cannot be trusted. Here's why: The sample only included 31% men, 10% voters under the age of 45, and 36% African Americans. Yes, the poll did weight the sample to make it resemble the actual electorate. But it had to "upweight" men by 50%, younger voters by 300%, and African Americans by 35%. Such a drastic adjustment means two things: First, we don't have an accurate sense about how men, younger voters, and African Americans are actually voting b/c their sample size is too small. Second, the pollsters were so lax in their "quotas" or "clusters" that they were clearly not particularly attuned to proper sampling in the first place.

The same poll had Obama up by 15 a couple of months ago. Like ARG, this is a very bad pollster whose results should not be trusted.

More oh-eight: A new SC poll

RH Oregon — Thu, 26 Jul 2007 13:32:39 GMT

Yes, but they will 'gleefully' quote it.
Just another soundbite.
Elect the House every 4 years.
Elect the Senate every 6 years.
Elect the President every 8 years.
Cut down on the Bull....!

More oh-eight: A new SC poll

Lyn, Silver Spring MD — Thu, 26 Jul 2007 13:52:01 GMT

Jon, all the statistician I know say the same thing. But I'm sure these guys know this or they have people smart enough to know this, so putting this out with No qualifiers is for a reason. The question is WHAT is that reason?

We know why Hillary did. They want to change the subject from yesterday. That don't want her Iraq voting record highlighted anymore. She has been successful so far keeping it under the radar. Attacking Obama was a big mistake. Time will tell how big?

More oh-eight: A new SC poll

Jane, Hoffman Estates IL — Thu, 26 Jul 2007 14:04:18 GMT

Polls. Polls had Howard Dean as the Democratic nominee in 2004, showed that Kerry won in 2004, and Gore won in 2000. Why anyone even bothers with them is beyond me.

More oh-eight: A new SC poll

TEC-SPRING-TX — Thu, 26 Jul 2007 14:08:22 GMT

RH Oregon, We have to many career politicians in office as it is. We should vote out everyone in the house every 4 years, restrict senators to 6 years and reelected a new president every 4 years.

More oh-eight: A new SC poll

jerry/corpus christi texas — Thu, 26 Jul 2007 14:15:02 GMT

probably 5000 polls out there, by the time you get around to examine each and every poll, the election will be over.

More oh-eight: A new SC poll

RH Oregon — Thu, 26 Jul 2007 14:27:49 GMT

No incumbunts.. Hmmm, I like it.
I would still not give the President less time in office than we give a senator.

More oh-eight: A new SC poll

Carrie, Eastern Iowa — Thu, 26 Jul 2007 15:07:13 GMT

Polls mean nothing at this stage anyway. Like Jane said, if we let the polls at this stage 4 years ago dictate who was the nominee, it would have been Howard Dean. They don't measure anything but the craftiness of the pollsters.

More oh-eight: A new SC poll

TEC-SPRING-TX — Thu, 26 Jul 2007 15:24:37 GMT

RH Oregon, do we really need a long term commander in Cheif? The president has limited powers as stated in the Constitution. Bush has exercised unprecedented powers thanks to the weak congressmen and senators. Congress has failed the American people, so they all need to go.

More oh-eight: A new SC poll

Jane, Hoffman Estates IL — Thu, 26 Jul 2007 16:21:08 GMT

Polls are no longer used to measure progress, or lack there of. It's not even clear that they ever were that unbiased. They are used by various camps as weapons against others. They are not used to measure peoples opinion, but to drive peoples opinion. Anyone that believes the results of a poll should proceed at their own risk. Poll results are nothing more then biased data points.

This goes for not only elections, but for the favorable/unfavorable ratings polls we see so much of now. People read a lot more substance into those numbers then there really is.

Only one poll matters, the one done on election day.

More oh-eight: A new SC poll

F StLouis — Thu, 26 Jul 2007 17:00:47 GMT

Let the polls continue. I like them. Lets just get destructive George out of office and be happy again. The man is a useless piece of garbage.

More oh-eight: A new SC poll

Jeff Morgan, Austin, TX — Thu, 26 Jul 2007 17:45:07 GMT

On polling, we of course have to be conscious of the levers that can be pulled in order to get certain responses.

A key question, such as choosing between Hillary or Obama, can be preceded by a series of questions intended to set up a particular answer for the key question.

Also the exact language matters. It matters if you ask 'which candidate will be better for the United States' VS 'which candidate will be better for the American people.' 'Who would you like to win the election' VS 'who would you like to be President.' It could be that certain phrases do better for a given candidate, and pollsters finding them means they have control over some of the percentage points. If they can get just a 5% impact they can turn a 10 point lead into 15. Do enough polling and those phrases will be found out.

Telephone polls also have a problem that over 60% of people called by a pollster hang up. Regardless of why this happens, that's the majority of people unseen.

Also polls can be 'soft.' For example think of a Joe Biden favorability scale from 1 to 10, and I'm Joe American taking a poll. I have a gray patchy fog floating over the region from 4 through 8. I really haven't made a strong evaluation; but have a good impression. Then I'm asked to pick a number, but it's arbitrary, could change on a whim. I pick 7. When my result is analyzed, you don't see that patchy fog, you see a distinct line. And then after a debate when I'm still equally foggy about Biden and give a 5 the following week, the pollsters will think it means something special when it's just that a thick patch of that fog was toward that 5.

When you see those poll percentages, remember the softness. We have no idea where preferences will be in the face of a serious factor that hardens people's view. We don't yet have any serious factors to test the candidates and solidify people's view (keep in mind most people aren't paying attention yet)

Polls will be in Hillary's favor for months to come; polls tell you where things are today, but what's important is what's to come tomorrow. Hillary loves the polls because today she is ahead. Obama's campaign is comfortable because of what their working on for the future.

More oh-eight: A new SC poll

RH Oregon — Thu, 26 Jul 2007 18:48:13 GMT

TEC - I wonder if the 4 year cycle gives any Presidents policies time to grow. Mainly I mean domestic policies, not foreign. I am of the opinion that the American economic engine is a slow moving beast. I personally believe that the balanced budget that President Clinton enjoyed for a short time was the result of 12 years of Reaganomics. It took time for the taxes from a better business climate to reach Washington. If Clinton economics created the balance why were we 2 months from declaring a recession when he left office?
I still like the idea of no incumbants.

More oh-eight: A new SC poll

Yiannis, Baltimore, MD — Thu, 26 Jul 2007 19:02:25 GMT

Carrie, FYI, four years ago this month, Lieberman was leading the polls, Gephardt was second and Kerry third. Dean wasn't even in the teens.

I think it is absurd that polls with sound methodology that are accepted across the board are not highlighted but instead everything is bunched together.

More oh-eight: A new SC poll

Matt Towery — Thu, 26 Jul 2007 21:35:41 GMT

I read with interest the attack on the sample and methodolgy of our poll. As we all know, every polling firm is vulnerable to criticism. We are used to that. Rather than debate this particular survey, let me ask the individual attacking our survey one simple question. Why, if our sampling and methodology is so wrong, have we consistently polled virtually every political race in our seven years of existence correctly (as to the winner)? In fact, we were the only firm to accurately poll the outcome of the Iowa caucus (Dem) four years ago (see Robert Novak's column days prior to the vote). I can certainly see ways to argue that certain groups are "undersampled" but anyone and everyone knows that undersampling occurs in virtually every poll (more women than men by usually a large margin). The fact is the survey is right in line with other respected pollsters as to the SC primary. As to why we had Obama up several months ago--I think I answered that in the Southern Political Report. The fact is that in southern contests the African-American vote usually has a high undecided until the very end. Our surveys make clear that "no opinion" is a choice. That's critical or you force a false result prior to groups having made up their mind. The attention this contest has received in South Carolina has led to an early consolidation of that vote, and in that consolidation Sen. Clinton has managed to split the African-American vote. Our company is owned by the owners of a billion dollar plus newspaper chain and one of the largest real estate investors in America. We are not allowed to be partisan (sorry--no "R") and are charged with getting it right. Most firms are dilligent in trying to do so as well, which is why I rarely criticize the work of others. I hope this adds some flavor, if nothing else, to this posting! Matt Towery, InsiderAdvantage

More oh-eight: A new SC poll

H P Boston — Fri, 27 Jul 2007 02:59:24 GMT

MMMM if the poll results were Obama 43, Clinton, 27/28
I guess you guys would be having a come to jesus party! All the posts would be yelling it must be 27% why are they saying 28%....Dang First Read you love Hillary! Yes I do think if the shoe were on the other foot you guys be LOVIN AND BELIEVING THE POLLS!