Redstate

"Get out. Vote. Donate. Volunteer. Help change the game."

Exactly .... We dont just witness history, we MAKE IT through out actions.

This is great to hear, these polls, it seems that GOP will take some inevitable hits, but will at least have a chance to ...

On another thread I was making the case that we shouldnt count DeWine and Santorum out ... I said:

An 8% gap looks bad but reality is that Sherrod Brown is a leftwing extremist who deserves to lose that race. Most voters only dimly know this 2 weeks out, and get their information from the 'media noise'. As they focus and pay attention, the GOP perks up ... IF THEY RUN A GOOD ISSUES ORIENTED CAMPAIGN THAT SHARPENS THE DIFFERENCES AND MAKES THE CHOICE CLEAR.

Santorum is running a good campaign, with some good ads ... See:
http://www.ricksantorum.com/uvc/

Rick Santorum deserves to win.

It's nice to see all of this info in one place like this. And the best news is that the swing back toward the GOP in these key races is before a coordinated response from the national level -- if there is even one coming.

The sad thing is the MSMedia reports will still be doom and gloom until election day. Good thing we've got the new media like you guys.

--------
Sometimes the hardest thing and the right thing are the same. -The Fray, "All At Once"

I don't believe them when we're down, so I can't very well believe them when we're up.

But here's hoping that this is a morale boost for others, and that it leads to increased effort and a better chance of victory!
--
If you're seeing shades of gray, it's because you're not looking close enough to see the black and white dots.

I'm not into the conspiracy theories on polls and some secret backroom that is plotting ways to make parties look bad. But I do think polls are an inaccurate science, especially when it comes to close races.

While it does give you a good gauge in races that aren't close (if someone is leading by 20% in the polls, I'm going to assume they are going to win the election), I don't think it accurately can portray any close election.

First, we've seen over the years that Republicans have been better at getting out to vote than Democrats. In a way, I've always viewed Democrats as those that talked about how bad things were, but just didn't have the time (or care) to actually do something about it. Whether this is a product of the social demographics of the voting party, or simply a lack of giving the voters a reason to head out to the polls, it's been clear over the past few elections that Republicans have done a better job of actually following through with what they believe.

The second issue I have with polls is whether someone will actually give you an honest answer. I'm sure most people do, but sitting in a private booth is much different than telling someone directly over the phone what your intentions are.

Finally, you are still dealing with small sample sizes. When 2 million people vote in a race and you poll 1000 people, I can't fathom how you can not deal with statistical anamolies, inaccuracies, and so on.

I think we need to go back to calling polls what they are, a guess. Stop trying to turn polls into a science and saying that because 500 people you called in the world tells you one thing, doesn't mean the other 300 million feel the same way.

I don't think most of the public polls are intentionally getting it wrong. I believe the problem is systemic.

Response rates - the % of adults reached who will consent to an interview - have fallen from about 80% 25 years ago to so low today none of the polls will even discuss it. Even those who disclose "internals" consider the response rate "proprietary information" and keep it secret nowadays, although indications are it is now barely over 20%.

If true, what that would mean is for a sample to be equally valid as one taken in the 1980s, the poll would have to make four times as many calls. Since the calling is done by humans, that means an increase of 400% in man-hours to get the same sample.

I suspect most of them simply take what they can get instead. So, while they could screen "hard" for likely voters by asking whether they voted in 2004 and 2002, now they simply accept the respondents' declarations that they will vote.

Add to this the proliferation of cell phones, the increased aversion to telephone solicitation generally as evidenced by "do not call lists," and the suspicion of polls by many, especially conservatives, and getting an accurate sample just costs too much for someone who is basically doing it as a promotion for their private polling services where they actually make money.

Internal party polls can be constructed to be more accurate, if the candidate/party is willing to pay the higher price to get it - but these are seldom disclosed - NEVER if they aren't favorable.

The deficiency of samples is confirmed, IMO, by the wildly fluctuating results in the same race from poll to poll, often even by the same pollster.

Garbage in, garbage out.

At the same time as SUSA, Mason Dixon said it's McCaskill +3. Somebody'd off big time. What gives?

Sitting out is a vote for KOS.

I meant "somebody's off". Sorry.

Sitting out is a vote for KOS.

"Our present addiction to pollsters and forecasters is a symptom of our chronic uncertainty about the future. Even when the forecasts prove wrong, we still go on asking for them. We watch our experts read the entrails of statistical tables and graphs the way the ancients watched their soothsayers read the entrails of a chicken." -- Eric Hoffer

also affected by the diminishing use of telephone land lines? More and more people use cell phones and broadband internet access exclusively and these numbers are not in the phone book.

Can you point me to the GA-11 poll?

Oz

Very latest has both pulling comfortably ahead. That's it folks.

Though, bad, DeWine still has big problems in OH.

Eric Dondero
www.mainstreamlibertarian.com

I'm more of a believer in the theory that the media is trying to suppress the Republican vote by consistently trumpeting the polls from 3 weeks out of an election cycle, which have been hostile to Republicans for the last three election cycles.

This is accurate. Think in terms of issues: often what doesn't get polled is important, too (late term abortion, amnesty, tax hikes, and so forth) and reveals a bias. Also contrast polls that show Republican gains--they aren't treated the same, to put it mildly, in the MSM.

I wonder if the advent of cell phones, telemarketing, caller ID, and other communications advances have destroyed the old methods. Obviously, the polling companies want to get things right (even if the MSM doesn't) and there has to be some reason polls show trends more than actual results these days. I always get frightened when people say "the only polls that matter are those on election day" and similar cliches, but in the last three or four cycles those trite and scared remarks have proved out.

The trend in the last week has been with the GOP; how that translates into individual races is hard to discern.

So the LA Times with their leftist bias has Talent and Corker increasing their leads. This is critical. I also don't buy that in the last 10 polls, only this LA one has Allen trailing. He will win in Virginia. If we could get one more out of MT, MD, or NJ/MI it would be a miracle. I could live with 51 or 52 seats.

United States Air Force
Cross Into the Blue

The issue of large MOEs is even worse than you imagine, but I think critics are missing this story by focusing on bias alone, as in, “polls oversample Democrats". Your post hints at the “other” problem, which is the excessive variation in poll results. It makes no sense to say a poll has a margin of error of, say, 3% if successive polls by even the same pollster bounce around by 5-10% or so – unless one also believes that the electorate is somehow ahistorically volatile.

Statistically, there aren’t too many ways to account for the generic problem of excessive volatility in the sample estimate of rates. In other contexts (bond default rates, mortgage loan prepayment rates) the culprit ends up being correlation in the sample. Though I can’t definitively lay out the mechanism by which correlation enters political poll samples, I suspect that some of the gymnastics and shortcuts that harried pollsters under deadline might go through to get any sample at all (say, purely hypothetically, polling groups at the shopping mall or more than one family member in a household) might inadvertently introduce the dreaded correlation demon into the sample. And it doesn’t take much to cause mischief: even a correlation of as little as 1% between poll responses is enough to blow up the margin of error to more than 10%, regardless of the size of the poll sample.

Bottom line: On this admittedly theoretical basis, I always assume the margin of error in poll results is at least 10% regardless of the poll’s stated precision. This policy has saved me a lot of head scratching when reading poll results.

If I'm pressed for time, I politely tell the tele-pollster to buzz off. Who knows what this phenomenon may be doing to the numbers? Are Republicans more pressed for time than Democrats?

Sincerely,
Bob Miller
Indianapolis, IN