Why Our Recent Polling In PA (Is Accurate) and Shows a Close Election for President

Since there is a consistent, multi-decade trend of differences between turnout in presidential years and off year elections, it’s quite interesting that you think this will look like an off year election, breaking a decades long trend. I can understand it looking like ’04 instead of ’08 in percentages of Democrats, minorities, etc. turning out, but ’10? I see no emperical basis for this judgement. Second, you didn’t mention a key factor, the manner in which you polled. If you polled landlines only, then there is a documented republican bias.

No poll should be based off of any turnout model…. Most polls do not assume any particular turnout model.

What shocking to blunt a week full of stories about their candidate losing momentum and falling behind — the GOP happens to pay for a poll, which just happens to find that a race which even Rassmussen has as Obama +12, is really a 1 point race?

Do people really take this poll or this company seriously? Nobody weighs the ages of a poll by either being under 44 or over 44… that’s such a nonsense methodology I’d be shocked if they actually called a single person.

And then there’s the University of Colorado economic model that has correctly picked the winner of the presidential race since 1990. They are going with Romney: http://bit.ly/NkEZHp

I disagree. Real statisticians would understand the “Likely Voter” polls already account for turnout, and thus to arbitrarily decide which party is more “enthusiastic” and will “turn out” renders a poll too subjective to be taken seriously. I think it’s worth nothing that Nate Silver, who called every state except Indiana in 2008 (which Obama won by less than .5%) and every senate race in 2008 and 2010 correctly, sees an Obama win in PA as a 95% certainty. His model, which is the gold standard to unbiased people who seek the truth, considers not only poll numbers but economic indicators and previous polling trends dating back decades. Learn more: http://www.fivethirtyeight.com

This election is a pure turnout battle – much like 2004. That is why I don’t believe ANY polls, and this is why the MSM is trying so damm hard to depress Republican turnout by convincing voters that Romney has no chance anyway so why vote? No one really knows what the turnout will be like, but my expectation is that the youth vote will be way down since Obama did nothing for them (I agree with the author on this), but black turnout will actually be UP from 2008. Blacks see Obama as an embattled member of the Tribe, and they will turn out 100% to support him for pure tribal solidarity. The big open questions are white turnout and Hispanic turnout. I really have no idea how this shakes out, since the MSM has immense power to shape the narrative and depress turnout of GOP leaning voters.

Here are the National percentages for 18-29 year olds:

(http://www.ropercenter.uconn.edu/elections/how_groups_voted/voted_08.html)

1980: 23
1984: 23
1988: 20
1992: 21
1996: 17
2000: 17
2004: 17
2008: 18
2012 (Susquehanna Projection): 7

Pay close attention to the paragraph where the pollster attempts to explain this absurd projection.

“For instance, according to 2008 exit polls voter turnout among 18-29 year olds peaked at 18%, but national and state polling proves interest among younger voters down sharply this year due to higher unemployment with younger voters and college graduates in particular. So conducting approximately ten percent of surveys with 18-29 year olds is a reflection of this lower anticipated turnout among these less-enthusiastic voters. Besides, the fact that Obama backers have suggested that over sampling older voters skews results in favor of Mitt Romney is a striking revelation in a state like Pennsylvania known for having the 5th largest population of senior citizens in the country.”

It implies that 2008 was an anomaly that “peaked” when it was actually right within the historical norm and even on the low side. Exact numbers are used throughout the article but somehow 7% turns into “approximately ten percent” in this context. 7% is an absurdly low projection. The truth is that the relative number of young voters hasn’t changed much over time and it is unlikely to change much this year. The author doesn’t base his numbers on any actual polling. Instead it is all based on his apparent personal opinion that higher unemployment is going to cause young people to stay home in droves. There are just so many specious assumptions that go into this theory. First it assumes that the same person is less likely to vote if he is unemployed than if he has a job. No evidence is given to back this assumption. There are several reasons to believe the exact opposite. Second, it assumes that the unemployment rate of young people relative to older people is considerably different than it was 2008. Again, I don’t have the numbers in front of me but I don’t think this is true. If unemployment is a serious problem for turnout projections, then presumably it would also affect keep older unemployed workers home. Unless, there was a dramatic change in the relative number of young unemployed workers, it shouldn’t affect the relative number of young voters. Third, it forgets about the fact that the vast majority of people are not unemployed. Remember the change is not particularly dramatic.

The current unemployment rate is 1.5% higher than it was in 2008. Lets assume for the sake of argument that unemployment has been a much bigger problem among the young than among the old. I’ll say that unemployment has gone up by 3% among younger voters and 1% among older voters. I’ll also assume that unemployed people are 20% less likely to vote. That means that the net effect is that relative youth turnout should drop by 4/10 of one percent ((.03-.01)*.2). The pollster appears to be assuming that the same effect will cause a drop in relative youth turnout by over 61% (.07/.18)

There is just no evidence whatsoever that there will be any significant drop in youth turnout. He also appeared to significantly underestimate minorities and urban areas. This is really amateur hour. It appears that the poll sponsor wanted a pro-Romney result and was willing to fiddle with the demographics as much as needed to get to the bottom line result he wanted. This is a textbook case of changing the evidence to match the conclusion instead of using the evidence to reach a conclusion.

On a macro level, I think people might be a little bit quick to write Romney’s political obituary. He is clearly very far behind, but he could catch up. Pennsylvania is probably out of reach but other states aren’t. If you are a Romney supporter, you need to be realistic. You still have a chance but it is a small one and making up numbers won’t improve your chances at all.

Still, the bottom line is that pollsters are supposed to be neutral and use realistic assumptions. This pollster is clearly not neutral. He can’t accept the fact that young people are telling him they are planning on voting for Obama so he is making up absurd reasons why they won’t vote at all. When the youth turnout eventually comes in the same narrow window it always does, I think he should be forced to explain exactly why he was so confident that they would be so severely underrepresented.