A response to a believer in the skewed party identification logic.

[Hyper_Eye]

On Facebook I noticed a conversation occurring in which someone stated that the polls are oversampling dems by basing their turnout expectation on 2008's turnout. They essentially said that if dems come out at +8 then the polls will hold up. Otherwise Obama is toast he said. Of course we all know that this is absurd. There are a lot of polling firms at work right now using finely honed scientific models to comb through data from thousands of respondents. They know a lot more about poll conducting than any fly-by couch pundit. Here is my response to them on Facebook:

Ahhh... the skewed party identification argument against polling consensus. That is the theory that spawned the greatly entertaining unskewedpolls.com which I am personally a big fan of. Whenever I feel down and need a good laugh it is a toss-up between that and conservapedia to deliver the great laugh I need to enjoy the rest of the day. The problem with that theory is that party identification does not tend to be a reliable constant. It changes with the political winds and studies have shown that people that aren't active consumers of political news can change their answer on that question over very short periods of time. Even Gallup has noted the unreliability of party identification while stating that it is a poor barometer of poll accuracy: http://pollingmatters.gallup.com/2012/09/the-recurring-and-misleading-focus-on.html

Regardless of what Gallup may have been showing in the national polling, an outlier mind you and outlier Gallup polls have historically been way off the mark, polls have never shown a lead for Romney, in this election cycle, in the states where the election will be decided. The aggregate of polls for Ohio, Wisconsin, Ohio, Iowa, Ohio, Nevada, OHIO have never relented Obama's lead. Even given some close numbers in Ohio at times consistency and volume increases confidence and Ohio has an abundance of those. Even the right-leaning Rasmussen (which is not the most accurate poll there is no matter how many times right-wing pundits repeat it) does not show a Romney lead there and only did once in the last month. If the early voting numbers prove to be accurate these polls look pretty good. Also, Nate Silver pointed out what we see every election and that is the convergence of polls close to election day. That isn't just the national polling numbers (RCP is showing the national aggregate has swapped again btw which shows a trend towards Nate's popular vote forecast) but also the national polls catching up to what the statewide polling has been showing all along. The fact is that Ohio votes count more than the rest of ours and it has maintained this pesky floor for Obama that Romney simply hasn't been able to break through. Finding those EV's elsewhere is tough. Wisconsin is a reach and Pennsylvania is always a tease for R's but they haven't taken it since 1988. Throwing a couple ads their way a week out isn't going to be enough (and the real reason they are doing that is because they have cash and the rest of the markets are sold out.) The numbers look good there for Obama and they are probably going to hold. Give Romney North Carolina, Florida, Virginia (though I have never thought it would not be blue on election night), maybe Colorado and New Hampshire, even Iowa (yeah right!) and it still doesn't matter... it's OHIO OHIO OHIO. They get to cast the magic votes this election... lucky them.

Of course there could be an upset and all this data could turn out to be wrong. It wasn't in 2004, which resembled this election quite a bit, and it certainly wasn't in 2008. Nate's model even predicted a huge victory for R's in 2010 very early. But maybe Romney has a perfect night. Polls have a margin of error after all and the margins could all be strangely pulling in Obama's favor. Also Nate's model does not produce a certain prediction but a probability of outcome and the simulation currently produces a Romney win 13.7% of the time. Maybe the bias is there and the methodology needs to be re-evaluated. The people putting money on it at intrade don't seem to think so. But just maybe. I'll be watching for the possibility that the unskewed guys have it right. It would be quite a different night than people are currently expecting. If that is the case though here is the map they say you're going to see (LOL): http://www.unskewedpolls.com/unskewed_projection_2012%20president_02.cfm