Obama’s National Lead Based Entirely on Over-Sampling Democrats

I went to great lengths to debunk the over-sampling of Democrats in polling used across polling outfits.  As a brief reminder, they use the prior election as their base model (7 percentage points more Democrats than Republicans) for consistency sake but not for accuracy sake.  Only Rasmussen Reports aggressively surveys the public to get a sense of current party affiliation and weights his polls accordingly.  This is why, despite hysterical protests from the Left, Rasmussen consistently calls the Presidential races better than his competitors. One of the commenters, greymarch, mentioned some good work by @NumbersMuncher showed the +3.1% lead for Obama in the current (September 15) Real Clear Politics average of national polls was based on polls where Democrats were being oversampled by on average +6.1%. Now we have this disparity in graphic detail:

As you can see the X-Axis is the % over-sampling either way: movement to the left is Republican over-sampling while movement to the right is Democrat over-sampling. The Y-Axis is the attendant Obama lead which loosely correlates to how greatly Democrats are over-sampled. The real take-away which I have mentioned the times I blog national polls is that many of those national polls are HORRIBLE for Obama, namely the ABC/Washington Post and CBS/New York Times polls where you have large Democrat over-samplings but rather small leads for Obama. This means if Obama doesn’t meet or beat his stellar 2008 turnout advantage he’s in for a drubbing on election day.

These over-samplings serve a few purposes but mainly drive down enthusiasm for Republicans while assisting the Obama campaign with “bandwagon” supporters who simply like being on the winning team (they’re real and they count). If pollsters in conjunction with the Obama campaign create a negative feedback loop for Republicans such that the marginal voter doesn’t show up (definitely a well documented top priority for the Obama campaign) and assist with the bandwagon voter — a small but meaningful voter in close elections — then Obama can create the perfect storm he needs to eek out a close victory following one of the worst four-year performances for any President in modern times (Carter is the only arguable comparable).

That is the what and why pollsters are doing the massive Democrat over-sampling this election cycle.

IMPORTANT UPDATE: As mentioned in the comments section, Rasmussen Premium subscribers know that Rasmussen Reports polls are weighted D +1, not not D -2 (or R+2 however you like it) like the original chart posted.  @NumbersMucher has a chart reflecting this corrected information and the chart above reflects the corrected data.

16 Comments

  1. Posted September 16, 2012 at 1:12 am | Permalink | Reply

    Great site Keith 🙂 Here is something interesting I was looking at in Gallup today. http://www.gallup.com/poll/154559/US-Presidential-Election-Center.aspx

    Comparing the latest demographic data in this election with the final week of Gallup in 2008 which showed Obama up 13 among registered voters and up 11 with likely voters here is what I see:

    First, these numbers are HORRIBLE for Obama. Check out these groups that he is down in compared to 2008

    Men -7
    Women -5
    White -6 (Makes sense why Chris Matthews and MSNBC is spouting racism to guilt whites into voting for Obama. However, this is still by far the biggest ethnic group in the U.S. Obama is only getting 39%. That is big trouble)
    Protestant -6
    Catholics -4
    Indies -2

    Obama is actually polling roughly as well as he had in 2008 with youths and hispanics. However, there is a 20(!)+ point drop in enthusiasm when you look at the rest of the age groups. Also, there is a 10+ point drop in enthusiasm with income earners earning less than $36,000 where Obama polls strongly in as well. Republicans are 5 points more enthusiastic to vote than Democrats.

    Regionally, the biggest change has been in the Midwest, West and South. Romney is crushing Obama in the South and the Midwest is basically even where Obama has lost 6 points of support and 10(!) points of support in the West.

    There is NO WAY Obama is getting re-elected with these numbers. That is why Axelrod needled Gallup to tweak their methodology and why media pundits are griping about Rasmussen. Obama has nothing to run on so in order for him to have any chance. He can’t gin up his base to come out and vote for him. They need to have these pollsters use the 2008 turnout model to try to depress Republican voters and the conservative base by showing Obama in the lead or the race tied. The only way that I can see Obama can win is by manipulating the GOP voter’s emotions to depress them in such a way that enough numbers will stay home and he’ll squeak in.

    • ed
      Posted September 16, 2012 at 11:07 am | Permalink | Reply

      thank you so much for that report. Anecdotally, I have not met one Obama voter who will vote for him again. I’m an attorney in the medical malpractice field and my informal poll of MDs and Lawyers is pretty dismal for Obama. That’s good for the country.

  2. sunflower
    Posted September 16, 2012 at 3:15 pm | Permalink | Reply

    hey keith, thanks for the hard work and great snapshot of these pollls.

    re: “Only Rasmussen Reports aggressively surveys the public to get a sense of current party affiliation and weights his polls accordingly.” where do you get this info? it’s my understading that he reports how the party id split pans out each month, he doesn’t create it. gallop does this as well, reporting how their numbers average out each month & year, not creating/weighting for them.

    thanks!

  3. firebird4ever
    Posted September 16, 2012 at 5:38 pm | Permalink | Reply

    Rasmussen looks to be using a weight of R+1 by averaging out his partisan voter index. In August he showed the GOP with a 37-33 registration advantage which is the highest recorded in Rasmussen’s history. If he was using an R+4 weight, Romney would probably be up by 5-6. I think it is too unrealistic to think the turnout will be R+4 election night. I think we’ll probably see something like we saw in 2004 and 2010. In 2004 it was 37D/37R/26I. In 2010 it was 35D/35R/30I. The amazing thing about 2010’s election was Democratic voters showed up to vote which I really had to hand it to the Democratic Party for turning out there vote in such an awful climate created by the passage of Obamacare. The GOP gained 63 seats in the House and it could have been another 20 if Democrats had stayed home. Gallup and other nationally known pollsters using D+7 (2008 turnout) or greater advantage towards Democrats this election isn’t going to happen. Indies have been peculiar in this presidential race. They are all over the place.

  4. Crowfoot
    Posted September 17, 2012 at 6:54 am | Permalink | Reply

    If you draw a line through the points on that graph, a Dem sampling advantage of +3 seems to result in a tie between Obama and Romney. And as others have mentioned, the average turnout advantage for Democrats in the last five elections has been 3.6. So from that we might conclude the candidates are more or less tied at the moment, perhaps with a slight edge to Obama.

    But actually, I think there’s a mistake in that graph. Rasmussen, despite what a lot of people think, isn’t using a GOP +2 sample in his poll results. For the last month, he’s consistenly used a Dem +1 sampling (or weighting, I imagine). You need to be a Platinum subscriber to Rasmussen to see that, but that’s what he’s doing. For the week of Sept. 3-9, he had a D/R/I weighting of 37/36/26 that had Romney up 47-45. The week before was 37/36/27 and again it had Romney up 47-45.

    So if you plot Rasmussen at Dem +1 (instead of GOP +2) and draw a new line, it looks like a Dem sampling advantage of about 4 corresponds to a Romney-Obama tie. Which means if we use the historic average on turnout day (i.e., Dem +3.6), Romney’s slightly ahead at the moment.

    • Posted September 17, 2012 at 12:56 pm | Permalink | Reply

      Thanks for the heads-up. I tracked @NumbersMuncher down and he already had a new chart ready reflecting the corrected data. Very much appreciated on the help!

  5. Marbles
    Posted September 17, 2012 at 2:02 pm | Permalink | Reply

    I think the crucial thing here is that most polls aren’t weighting their samples or oversampling. They’re calling on a random cohort who seem to be identifying as democratic (this isn’t necessarily what they’re registered as, just what they say on the phone on the day). This has no effect on likely voter models, which are generally based on enthusiasm. The main reason more people are identifying as democrats? Because more people plan on voting democratic.

    • David Weed
      Posted September 30, 2012 at 1:05 pm | Permalink | Reply

      Amen, Marbles. You’ve lost none of them.

  6. Gayle
    Posted September 20, 2012 at 3:54 am | Permalink | Reply

    yes, and also when they rig the election, it won’t be noticed if they skew it to look like obummer is ahead …I was told they can rig it is within ten points.. No doubt in my mind he will cheat in many ways..some like they did in 2008 and some new ones such as Soros buying company that tallys election votes…hmm…..don’t sound good to me..we know these are corrupt people and have no honor

  7. Marian Tanenbaum
    Posted September 25, 2012 at 9:17 am | Permalink | Reply

    Most recent analysis of the state of the election assumes poll validity. But social desirability bias, i.e., the Wilder effect, and sampling bias, both intentional and unintentional, render the polls all but useless. (The leftist media is, of course, the source of the intentional bias. Pollsters’ inability to accurately predict turnout is the primary source of the unintentional bias.)

    Moreover, the dramatic collapse of the Obama presidency on all fronts and in full view magnifies the biases, especially the Wilder effect.

  8. Bill
    Posted September 28, 2012 at 10:25 pm | Permalink | Reply

    I think you embarrass yourself with your post. You apparently don’t understand how polling is done.

    In the article below, Gallup basically says that party affiliation follows support for a candidate. Thus, if they call 500 people and 300 of them say they support Obama, it is more than likely that those 300, if pressed will identify as Democrats. That doesn’t mean that Gallup called up 300 Democrats and only 200 Republicans, as you seem to think. It’s simply a reflection of their polling that relies, not on party affiliation, but on “variables such as age, gender, race/ethnicity, and education.” Perhaps a little more scholarship on your part would make your claim less absurd?

    http://pollingmatters.gallup.com/

    • Posted September 28, 2012 at 10:34 pm | Permalink | Reply

      Bill, thank you for visiting and embarrassing yourself in your very first comment. A truly remarkable achievement. The gallup defense is laughable. He shot his whole argument at the end with the following: “So, if one sees a poll saying that Obama is leading Romney by nine points in Florida, then one should ask how likely it is that Obama will exceed his 2008 margin by six points. That is a reasonable discussion. But one need not attempt to say that the nine-point lead in the poll is suspect because there were too many Democrats and not enough Republicans in the sample compared to 2008.” That is a foolish and ludicrous statement not worthy of even Internet tripe let alone under the pen of the Gallup organization. Our entire argument is you are seeing Obama with a 9-point lead BECAUSE you have too many Democrats. He has the tail wagging the dog by saying Obama has a 9-point lead because more people identify with Democrats. If believes that, then fine. But give me an argument for why more Democrats identify with Obama today versus 2008 when he then had countless advantages but today he has countless baggage. He totally dodged the subject and misrepresented our complaint.

      The 2008 election was a banner year for candidate Obama and Democrats in general. The top of the ticket was a historic candidate (Obama), America had war and Bush fatigue, the financial meltdown created an anti-Republican wave, and his opponent wasn’t the strongest (good biography, bad candidate). These factors led to a strong Democrat advantage at the voting booth where 7% more voters identified themselves with Democrats than they did with Republicans (39 to 32).

      But in the 2012 election, none of the advantages outlined above are there for Obama and many of those factors are now largely working against Obama: 8%+ unemployment for three years, sub-2% GDP, 23 million unemployed, Arab Spring blowing up and casting the historic vote in 2008 is yesterday’s news. Additionally the Romney campaign ground game has exceeded the McCain campaign across many metrics as much as 10- to 15-fold.

      Despite the stark changes in each of these factors, polling outfits thus far have consistently sampled an election turnout often greater than candidate Obama’s 2008 best-in-a-generation advantage.

      That means something else is going on. But the polling organizations shrug their shoulders and say the losers in the results are just crying sour grapes. This is even though their sample sizes are unrealistic beyond any stretch of reason. If they believe today’s electorate will exceed the incredible 2008 advantage Obama achieved they should say so. But they would also have to explain why the 2008 election gave Democrats massive majorities in the House of Representatives yet today’s electorate will return massive majorities in the House to Republicans. It defies all logic or reason. But I’m guessing due to “confirmation bias” the media and polling organizations don’t doubt their own numbers. That is, because they want Obama to win they report these numbers as accurate when clearly they are not.

  9. David Weed
    Posted September 30, 2012 at 11:00 am | Permalink | Reply

    Regarding Bill’s post (#8) and Keith’s reply. I’ve read the Gallup article twice now. The arguments are compelling. Nate Silver posted an article today (Sunday) mirroring the exact point, while also providing historical data. Now, you may not agree with Nate Silver’s 538 analysis. All of those pesky unfriendly numbers. Certainly Silver’s predictions are unfavorable to someone who supports Mitt Romney in the Presidential election.

    The question posed by Gallup that Keith derides is actually quite a good one, and it’s a question that this site tends to overlook. Gallup is asking, if a poll favors Obama by a larger amount than the President won in 2008, WHY this might happen.

    The poll numbers have shifted steadily for Obama since the conventions. Is the reason for these numbers oversampling of Dems, or have events influenced people polled to identify themselves as more likely to vote for the Democrats? The polls were very close at the end of August.

    1. The DNC had the advantage of going second in the conventions, and right after the GOP convention. Most people agree that the Democratic convention was more successful than the Republican one. You might have loved Clint’s chair routine, but not everyone did. You may have hated Clinton’s speech, but it appeared to have resonated with voters. Do the readers on this site contend that the GOP convention was more successful than the Democrats? The polls begin to favor Obama at this point, though.

    2. Romney then jumps the gun late on 9/11 and the Libya consulate attacks, turning what could have been a positive (for his campaign) into at best a break-even moment. Mitt jumped the gun, before having all of the facts, in order to score political points. What should have been more of a story politically was muddied by Governor Romney’s actions. Instead of focussing SOLELY on the events, and how much President Obama should be held responsible, much of the debate was spent on whether or not Romney jumped the gun for political gain. Obama’s numbers continue to climb.

    3. As those rising numbers begin to steady, Politico runs a story on September 18th about dissension in the Romney campaign. However, later that day, the 47% tape is released by Mother Jones. Now, can ANYONE on this site tell me with a straight face that the fundraising tape wasn’t complete poison for the Romney campaign? That tape affirmed Mitt’s greatest liability- that he’s a wealthy man, FOR wealthy people, and is out of touch with the middle class. That May recording is DEVASTATING to the campaign. And we’ve seen a steady rise in President Obama’s numbers since.

    This site contends that Obama’s numbers are inflated because of oversampling of Dems. Gallup suggests that more people identify themselves with the Dems because of recent events. And that party

    The above events happened, people. Which is more likely? That nearly EVERY polling firm is screwing up, or that Mitt simply had a really, REALLY bad month, filled with gaffes, mistakes, and the damning 47% footage, which (to Dems) is the gift that keeps on giving.

    And when considering Obama’s 2008 numbers, please consider that Mitt’s favorables are CONSIDERABLY worse than John McCain’s. Yes, the economy is in bad shape, but not everyone blames Obama completely for it. Had the recession begun in 2009, the GOP could have run anyone and defeated Obama handily. But the collapse began in 2008. And President Obama is simply more likable than Mitt Romney. People simply do not like Mitt. Heck, many Republicans simply tolerate him. I don’t think it’s a stretch to say that a large percentage of GOP voters are voting LESS for Mitt and MORE against Obama.

    And please, if you’re going to AGAIN cite voter enthusiasm, back it up with some facts and figures. Democrats enthusiasm was tepid UNTIL the DNC. It’s the first event we really had in this election. Once the DNC began, Democratic enthusiasm caught up with the GOP. One could argue that enthusiasm on the left will peak just at the right time.

    I look forward to the first debate, though. And I’m a little nervous. How well will Mitt do? Will Obama stumble?

    – Dave

    • Posted September 30, 2012 at 11:52 am | Permalink | Reply

      Dave, I bolded the font, made the text size large, centered it and made it flash repeatedly on the screen.
      http://battlegroundwatch.com/2012/09/29/about-that-enthusiasm-gap-were-not-in-2008-any-more-dorothy/

      • David Weed
        Posted September 30, 2012 at 1:02 pm | Permalink

        Yes, Keith. You quoted snippet from a poll from USA Today, which sampled 1400 ADULTS nationwide. And USA Today, one of the most respected journalistic institutions in America, failed to show ANY real numbers. Maybe you found them. I looked, though.

        It’s USA Today, Keith. It’s a nationwide sample of ADULTS. Not RV. Not LV. And it was one snippet in a sea of graphs that mostly favor President Obama. Except that it’s USA TODAY!

        Now, if you show me an article from People Magazine about voter enthusiasm, we can talk. Or a dissertation on Obama’s waning youth support from Tiger Beat.

        On their best day, USA Today releases an occasional poll in conjunction with Gallup. Now, Gallup has recently fallen out of favor with many of the readers of this blog. But here are GALLUP’S numbers, including Swing States and Nationwide:
        http://www.gallup.com/poll/157547/democratic-enthusiasm-swells-swing-states-nationally.aspx

        You quoted a USA Today article as if it were gospel, Keith. You included an awesome clip from The Princess Bride, but it doesn’t make that report any more credible. If I’m missing some of the actual polling data from that article, and you have it, I’ll gladly look at it. But you can’t dismiss nearly every poll from reputable firms as misleading because of oversampling while then embracing one of the least scientific polls without losing credibility.

        You state, as if it were fact, that there’s an enthusiasm gap. You cite USA Today and their polling of 1400 adults as your credible source. No amount of centering and bolding and inserted videos can mask that you’re cherry-picking.

14 Trackbacks

  1. […] of his arguments ultimately rise or fall based on polling numbers that have been shown to be likely wholly unrealistic samplings of the public–always greatly in favor of President Obama. This is the straight forward argument that Obama […]

  2. […] keithbacker at Battleground Watch sums up the graph this way: The real take-away which I have mentioned the times I blog national […]

  3. […] as journalists. This is all because of the incontrovertible truth that in mid-September in a race no worse than tied against a President presiding over an abysmal economy and a foreign policy currently aflame with […]

  4. […] flattered so many people picked up on my post regarding polls oversampling Democrats accounting for Obama’s entire lead .  Adding to the discussion, The Winston Group has a […]

  5. […] an anonymous poll analyst, tweeted the following graph, garnering Internet attention, here and here. Some conclude this is further evidence prominent public opinion polls have oversampled […]

  6. […] an anonymous poll analyst, tweeted the following graph, garnering Internet attention, here and here. Some conclude this is further evidence prominent public opinion polls have oversampled […]

  7. […] keithbacker at Battleground Watch sums up the graph this […]

  8. By » Graph Shows Obama Lead IS Bogus on September 19, 2012 at 8:34 pm

    […] keithbacker at Battleground Watch sums up the graph this […]

  9. […] keithbacker at Battleground Watch sums up the graph this way: The real take-away which I have mentioned the times I blog national […]

  10. […] By one analysis, pollsters are oversampling Democrats by an average of 6.1%. […]

  11. By The evidence continues to mount. on September 26, 2012 at 6:39 pm

    […] […]

  12. […] Ross and Battleground Watch carry the same informative graph on the top American […]

  13. […] than Republicans.  This is not a new phenomenon as I outlined in mid-September in the post “Obama’s National Lead Based Entirely on Over-Sampling Democrats.”  Today’s ABC/Washington Post poll is the crowning achievement this cycle in […]

  14. […] decided to post a radio interview I did last month based on the blog post discussing the idea that Obama’s lead in polls was entirely due to over-sampling Democrats (September […]

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: