The Worst Poll in the World

It seems to me that there has been a ton of manipulation of many pollsters to make Obama’s standing appear stronger than it really is.  You know, like the NBC poll where they sampled 11% more Democrats than Republicans nationwide or the one where nearly half the sample consisted of Democrats…. in North Carolina!

I sort of thought that this sort of ridiculousness might go away once we got after Labor Day into the heart of campaign season.  I was wrong.

Most people know that Ohio is a crucial state in the upcoming presidential election.  NYT political statistician extraordinaire Nate Silver has declared it the state most likely to be the “tipping point” state that decides the election.  Well NBC just tried to poll Ohio, but instead wound up with egg on their face.

I’ll just give you a couple facts about Ohio and the poll before getting to the punchline.

1) Ohio is definitely a swing state… its electorate was +5 Republican in 2004.  It shifted dramatically to +3 Democrat in 2006 and +8 Democrat in 2008.  but in 2010, it shifted back to even.  Somewhere around even to +2 Democrat would seem to be a good assumption of turnout in 2012.

2) The NBC poll in question had Obama up 93%-5% among Democrats, Romney up 92%-4% among Republicans, and Romney up 44%-41% among independent voters.

So logically, Romney and Obama should be about even, right?  Obama might have a slight edge in party registration, but Romney has a slight lead among independent voters.  Both candidates are ahead by exactly 88% within their own parties.  So you’d expect an incredibly close result…. perhaps Obama might be ahead by a single point?


This poll from NBC has Obama up 7 points.  The reason for this is that they are using a sample consisting of 37% Democrats and just 28% Republicans.

In 2010, Republicans consisted of 36% of the electorate in OH.  Even in the worst Republican year in decades, 2008, Republicans consisted of 31% of the electorate.  So NBC is predicting that there will be about 170,000  fewer registered Republicans showing up to vote in OH as compared to 2008.  There’s just no support for such an insane assumption.

2 other NBC polls released today, in Florida and Virginia, were guilty of similar skewing of the electorate distribution polled to manipulate the results to make them more favorable to Obama (although not quite to the absurd degree seen in this poll), as are many other recent polls.  Interestingly, there are incredibly few polls out there that seem skewed in a pro-Romney direction.  Shocking, I know.

It is clear the media is going to try to setup the narrative that Obama is inevitable and going to win easily to lower Republican enthusiasm.  It is a smart strategy, because in a close election, a 1% drop in Republican turnout could make all the difference.  I may blog further on this theme if (when?) more ridiculous polls like this come out.  But bottom line:  this is going to be an incredibly close election, especially in OH and WI, the two states that I think will decide the presidency.  There’s very few undecideds out there, so there’s an incredibly high chance it will remain close up to election day and turnout is just as (if not more) crucial than how those last few undecided voters break.

Now is not the time to despair…. it is time to win.



  1. Excellent final point, Ryan.

    I was talking with a friend about this the other day. Thank you for putting it to words so I can just link to your work in future discussions!

  2. Randy in Richmond says:

    Interesting read Ryan. There has been talk here for quite a while on Virginia’s polling being out of whack. A Rasmussen poll out yesterday has Obama +1 here in Virginia.

    My local poll of bumper stickers and yard signs shows much less enthusiasm on the Obama front. In fact, so far, I have seen more ’08 Obama bumper stickers than ’12.

  3. “This poll from NBC has Obama up 7 points. The reason for this is that they are using a sample consisting of 37% Democrats and just 28% Republicans.”

    That’s 65%. Most samples add up to 100%

  4. I thought it was obvious but the remaining voters are independents

  5. I followed that! It could be some fun.

  6. Okay, now this is getting ridiculous. The new CBS/NYT Poll for CO, VA, and WI used completely ridiculous crosstabs, especially for VA.

    Their results were as follows:

    CO: Obama by 2, with R+1 sample
    WI: Obama by 6, with D+8 sample
    VA: Obama by 4, with D+11 sample

    Realistically, I would expect D+2 in CO, D+3 in WI, and D+1 in VA. If you normalize to get a more plausible sample, you get the following results.

    CO: Obama by 3
    WI: Obama by 3
    VA: Romney by 2

    So the headline of this poll was: “A Tight Race in Colorado; Obama Has the Edge in Virginia and Wisconsin.” In fact, the data they got showed tight races in all 3 states, with Romney actually ahead in the all-important state of VA.

    It certainly appears as though they didn’t like the idea of Romney ahead in VA, so they removed some Republicans from their sample to force the result they wanted to see; an electorate of only 24% Republicans in VA is unfathomable (was 33% in the terrible R year of 2008…. anything below that is manipulation)

  7. Disingenuous and dishonest are some pretty consistent threads these days from the left leaning 🙂