Sep 27 2008
Why Many Polls May Be Too Optimistic For Obama
Voter models are the essence of political polls. You take a sample of a few hundred or a few thousand people and predict how that sample can reflect 10s-100s of millions of people. If you are off by even a small fraction in your assumptions the bottom line could be off by 5, 10 or 20% (despite an MoE claim of a few points).
We have a perfect example of this in two Colorado Polls out recently. Â The first poll was commented on by our Reader MerlinOS2:
PPP just released a poll in Colorado which puts Obama up +7
Now what the issue is here is that the party split was
Dem 40
Rep 36
Ind 24However August voter registration number per the spreadsheet available from the Secretary of State show the registration breakdown is
Dem 30.6
Rep 34.8
Ind 34.5
Details on the poll in question can be found here.  Just this week American Research Group (ARG) also released a poll for Colorado (which is not used in the RCP poll of polls strangely). Its voter model was Dem 32%, Rep 35% and Ind 33%, very close to the ACTUAL voter registration levels noted by MerlinOS2.  The result: McCain 48%, Obama 45% – a McCain lead of +3%. (Note: this polls also shows McCain tied with women)
These polls were taken at basically the same time in the same state. But we can see how the voter model can really change the bottom line (a 10% difference).
With that hard example in mind, let’s look at a recent PA poll. This poll has Obama up only 4% with an margin of error of 4.5% – it is a statistical tie (and should mathematically be treated like one). But here is the kicker, the party sample is Dem 51%, Rep 38% and Ind/other 10%. Even with this much slant towards the dems Obama has a measely 4% lead? How can this be? How could a 12% difference in party identification turn into a 4% lead?
Democrat defections. Massive democrat defections. And this is not the only poll to show this, here is another poll from PA in the same timeframe:
19% of Democrats are voting Republican. 17% of Republicans are voting Democrat. This is a level of party defection not seen in other states, and may be a window into just how unhappy some Keystone voters are with news events in general, as much as they might be unhappy with their party’s candidate in particular. McCain holds 81% of Conservatives, Obama holds 86% of Liberals, but Obama is positioned to capture PA’s 21 vital Electoral Votes on the backs of Moderates, among whom he leads by 22 points today.
Losing 20% of Democrats but leading with moderates??? OK, that’s one theory I guess. BTW, this poll also has a 51% D, 38% R and 10% I sample. Â Obama leads by 6%, supposedly. The 51-38-10 model is representative of the current state registration statistics. But one has to wonder how accurate these rolls are (I have not updated my registration in two decades).
I have been scratching my head wondering why Obama and his liberal media allies have been going out of their way to play to the hard left? Obama needs the moderates, but his anger and lies and BS seem to be directed at the nutroots – why? Maybe we are seeing why in this poll. In a dark blue state with a massive sample difference between Dems and Reps and Inds one would think Obama would be 12% head (or better). Clearly something is going on the polls are actually obstructing.
Update: Beware those snap post debate polls – they are tilted:
The first problem is the numbers:
Fifty-one percent of those polled thought Obama did the better job in Friday night’s debate, while 38 percent said John McCain did better.
Buried way, way down at the bottom of the story — hopefully, one surmises, past the point where anyone would read — is the following:
The results may be favoring Obama simply because more Democrats than Republicans tuned in to the debate. Of the debate-watchers questioned in this poll, 41 percent of the respondents identified themselves as Democrats, 27 percent as Republicans and 30 percent as independents.
We know the nation is not 41, 27, 30 – so the poll and its so called result are garbage.
The polls have been odd. Conflicting results and all of the claiming margins are error at about $%.
I also wonder about the undecideds. Who are these people? They go up and down and in and out.
AJ,
This is coming up again and again. Perhaps the information on party affiliation is correct. If so (and it really is something along the lines of 41-27, Dem-Rep, I think we can all go home now and concede defeat. You are regularly having polls now showing Dems up 8, 10, 12 percent over Reps in party affiliation. Again, if true, it’s over and McCain has no chance. However, it would be absolutely unprecedented. There have just been a few articles written about this in the past week (sorry, no links, because I can’t remember the sources). I believe the maximum spread in an election year, based on exit polling, looking back over the past 35 years, is 4% (Dems over Reps). In 2006, a year when Dems made big gains at the ballot box, polls were regularly coming up with 7, 8, over 10% more Dems than Reps in affilication. Exit polling showed it to be around 3% on election day. Then again, I don’t think Rasmussen uses such heavily skewed samples and they show Obama up by 6% today. Go figure.
In an unrelated issue, how is it that Obama can seemingly get away with saying that he has never been in favor of meetings at the Presidential level with the leadership of Iran? This is as blatant of a lie I’ve seen in a political campaign and demonstrably so- from video of him saying it to old screen captures from his website. He lied outright on national television last night. Amazing.
AJ
I know this is unrelated to this post but wondering if you have any more info on what is going on in MO with Obama campaign trying to use law enforcement to silence critics. Just saw the press release from Mo Gov Blunt.
http://governor.mo.gov/cgi-bin/coranto/viewnews.cgi?id=EkkkVFulkpOzXqGMaj&style=Default+News+Style&tmpl=newsitem
The polling ratios should reflect the actual poll turnouts. In other words, of all voters casting a vote in the last two elections, if 52% were Republicans, 48% percent Democrats, and the remainder Independent or Other, then the polling should be the same.
It doesn’t matter what voter registrations are, that assumes that everyone registered votes. They dont. Younger voters who are more likely not to vote are generally registered Democrat.
An example on the other side is our troops. While they seem to overwhelmingly support McCain, less than 10% of the military actually casts a vote so their voice in polls is pretty much meaningless. They don’t vote so their opinion matters little.
Look at actual voter turnout party breakdown and use that. That is what Rasmussen does.
The reason for the positions Obama takes is simple.
He believes dems will vote dem rather than cross over or sit out.
What he is relying on is the SEIU/ACORN registration drives for new voter and most of them college kids.
He is aiming for the college kids to be the make or break here and a lot of his message is playing to the FaceBook/MySpace crowd.
But unless I miss my guess the seniors are going to be more of a problem for such a left wing positioning than his internal polling is revealing.
Cochino,
In addition to the lie about Iran, how about the lie that Dr. Kissinger supports his position on meeting the leader of Iran, something Dr Kissinger denied within the hour after the debate. Have you seen any comments about it?
You can also add to the lie about Iran during last night’s debate the lie that Obama supports Anti-missile defense, something that he is on record, including videotape, that he would cancel should he become president.
[…] AJ Strata takes a look at discrepancies in data from Colorado and Pennsylvania while pointing out the clear differences in polling versus actual party affiliation numbers that exist in states. […]