A Corrected Presidential Polling Average

The pollsters now claim to be surveying voters likely to cast ballots in the 2012 presidential election, but Democrats outnumber Republicans by between 5% to 11% in these polls. This partisan bias, of course, has no reflection in reality where Rasmussen’s polling of well over 10,000 Americans finds that voters are currently registering 35.4% Republican, 34% Democrat, and 30.5% Independent and GOP voters are more enthusiastic than Democrats.

In a post earlier in the week, I reviewed the Edison exit poll partisan distribution of voters for all the election cycles and, based upon the a GOP enthusiasm advantage superior to that in 2004 and a small migration of GOP voters becoming independents, assumed that the 2012 partisan distribution of voters will look very much like 2010 – 36% Democrat, 36% Republican and 28% Independent. Remember that the Edison exit poll itself generally over counts Democrats (ask President Kerry who won the Edison exit poll in 2004). Thus, my assumption of 2012 partisan distribution is conservatively more favorable to the democrats than the current distribution of registered voters.

What I propose to do here is to take the established likely voter polls listed in the RealClearPolitics presidential polling average which disclose their partisan weighting, adjust the poll results to conform with the more likely historical partisan weighting of 36% Democrat, 36% Republican and 28% Independent, then average those adjusted poll results. You can find my rough back-of-the-envelope methodology here. Note: This is only a rough corrective to show approximately what the polls ought to look like with the Democrat bias removed, not a prediction of the election results.

Here are 7 likely voter polls for this week excluding Rasmussen, for which I do not have the partisan weighting:

IBD/TIPP
Original Poll Results: Obama 46%, Romney 44%
Percentage Democrats outnumber Republicans: +5%
Corrected Results: Romney 47.5%, Obama 43.5%

Washington Post/ABC News
Original Poll Results: Obama 49%, Romney 48%
Percentage Democrats outnumber Republicans: +6%
Corrected Results: Romney 52.5%, Obama 47.5%

Reuters/Ipsos
Original Poll Results: Obama 48%, Romney 45%
Percentage Democrats outnumber Republicans: +9%
Corrected Results: Romney 49.5%, Obama 43.5%

Fox News/OD
Original Poll Results: Obama 48%, Romney 43%
Percentage Democrats outnumber Republicans: +6%
Corrected Results: Romney 46%, Obama 45%

PPP
Original Poll Results: Obama 50%, Romney 45%
Percentage Democrats outnumber Republicans: +11%
Corrected Results: Romney 51%, Obama 46%

CBS News/NYT
Original Poll Results: Obama 46%, Romney 45%
Percentage Democrats outnumber Republicans: +6%
Corrected Results: Romney 48%, Obama 43%

CNN/ORC
Original Poll Results: Obama 52%, Romney 46%
Percentage Democrats outnumber Republicans: +5%
Corrected Results: Romney 53%, Obama 45%

Note: This poll was so wildly skewed – finding that Romney led Independents by 14% and then reducing independents to less than 5% of the electorate – that my rough corrective methodology also ended up skewed, so I used this more fine tuned correction.

Average of Corrected Results: Romney 49.6%, Obama 44.8%

In sum, many of these pollsters are anticipating a replay of the 2008 election turnout and still cannot get Barack Obama above 50% support. In the far more likely case (as I assume) that the 2010 electorate is showing up to the polls in 2012, Romney has a comfortable lead. Even if the Democrats manage to get another percent or two of their vote out as in the Democratic year of 2006, Romney is still leading.

About these ads

About Bart DePalma

Bart DePalma is a solo country attorney practicing in the mountain town of Woodland Park, CO. Bart publishes the Citizen Pamphleteer blog and his new book Never Allow A Crisis To Go To Waste: Barack Obama and the Evolution of American Socialism was released December 2011.
This entry was posted in Uncategorized and tagged , . Bookmark the permalink.

40 Responses to A Corrected Presidential Polling Average

  1. Martin says:

    You do realize that Rasmussen’s use of a 100% landline poll for partisan ID is going to markedly skew their results, don’t you? 15,000 samples from a skewed population doesn’t improve the absolute accuracy, it just decreases the variance about a wrong answer. Don’t you think it’s a little odd that *every* other poll comes back with a very different result than Rasmussen on partisan ID? See, if it were just random sampling error and Rasmussen’s result were the true answer, the other polls would be distributed on both sides of Rasmussen’s answer. But they’re all skewed to one side. Now, you can try to argue, if you wish, that organizations as diverse as Gallup, Investors Business Daily and Reuters are somehow all intentionally skewing their polling samples to make Mitt look bad, but it would be a pretty wild theory…

    • Bart DePalma says:

      The possible skew of landline vs. various hybrids of cell and landline is indeed an ongoing debate.

      Let us assume a skew to landline-only polling and that skew being constant. The corrective is to compare Rasmussen’s partisan registration percentages before each election to the actual election results to determine what the current partisan registration percentages tell us about the 2012 election results.

      Datechguy did the work for us and the 2012 election looks basically the same or a bit better for the GOP than 2010. http://datechguyblog.com/2012/09/17/demoralized-as-hell-the-poll-the-media-isnt-talking-about-edition/

      I will be posting on this shortly.

  2. Shiloh says:

    Most likely the internal Romney polling is matching the “Dem” polling. That is why they are starting to panic.

    • Bart DePalma says:

      Have you noticed that the same folks who are providing you the massively Democrat biased polling are also claiming from the usual unnamed sources that the Romney campaign is in a state of “panic?”

      These people really need to report their in kind donations to the Obama campaign to the FEC.

  3. Shiloh says:

    I don’t have to be told that they’re panicking, I can see it.

  4. Mike says:

    Bart, Shiloh is a TROLL that goes around to several conservatives blogs spewing his liberal opinions…….

    • Bart DePalma says:

      I know Shiloh from a liberal blog where I provided my libertaran conservative point of view.

      So long as they are polite, everyone is welcome.

      • shiloh says:

        Let the record show the above Shiloh is not me. Also, I no longer post at logarchism and the only con blog I post at is Althouse.

        As always, Bart and I agree on one thing, “Novem­ber 2012 can­not come fast enough. Then you will see the actual will of the peo­ple at the bal­lot box.”

        http://www.logarchism.com/2011/07/23/obamas-road-to-reelection/comment-page-2/#comment-8560

        ciao

        • shiloh says:

          Correction:

          The only con blog I post at is Althouse … and now cp also. :)

        • Bart DePalma says:

          shiloh:

          Good to hear from you.

          Is this other person an imitator or stalker of yours?

          • shiloh says:

            Bart, have no clue. Maybe a lib at logarchism just having some fun. But imitation is the sincerest form of flattery :) although I’m not as “sincere” as my fellow “Shiloh.”

            Speaking of logarchism, peeked in today and discovered you’re an HP super user! Say its’ not true! :D

            http://www.huffingtonpost.com/social/Bart_DePalma?action=comments

            btw, you have a lot more friends than me … and comments as well.

            http://www.huffingtonpost.com/social/stokely

            When I first joined HP, it was mostly libs w/a few cons. But now I believe the con trolls may be in the majority. Arianna is smilin’.

            >

            Imagined skewed polls notwithstanding, you must know Romney has to be the worst presidential nominee since Dukakis, maybe since McGovern.

          • Bart DePalma says:

            shiloh:

            Romney is a decent candidate (and would look far better in a tank than Dukakis), but a very suspect conservative. When he wins, Romney will be on probation.

            In a perverse way, you Dems ought to hope that Romney wins and then governs like a RINO. The Tea Party barely tolerates the GOP establishment now and another RINO president would almost certainly result in a third party (which I might join).

          • Bart DePalma says:

            Posting at Huff Post is a guilty pleasure like throwing balls at a dunking booth on which the self important twit class president is sitting.

          • shiloh says:

            Bart seriously, do you think Nate Silver is skewing his scientific results? Rhetorical. All these political pollsters depend on their reputation to stay in business. Again, their bottom line is their bottom line! It will be interesting to watch Rasmussen as the election nears and how they fall in line w/other pollsters.

            btw, In “your” 2010 wave Rep election :) Rasmussen had an error of (6) pts in the CO senate race and (9) pts. in the NV senate race:

            Angle NV +3 ~ Oct. 30, 2010
            Reid won by +6
            >
            Buck CO +4 ~ Oct. 30, 2010
            Bennet won by +2

            Oops! Scotty Ras must have been ((( wishin’/hopin’/prayin’ ))) rather than doing scientific polling!

            Remember when Nate exposed Research 2000 as a fraud. As always, don’t bet against Nate.

          • Bart DePalma says:

            “Bart seriously, do you think Nate Silver is skewing his scientific results?”

            Nate has to know that his models are relying on very questionable polling. His posts trying to justify using polls with massive partisan skews have been very unconvincing. I would have thought after his debacle in 2010 underestimating the GOP gains by 100% that he would know better. Unless the polls on which he relies right themselves at the last minute, he will be embarrassed again.

            I have also repeatedly questioned his use of various economic factors over time in responses at 538. He has a bad habit of not normalizing them for things like population growth. For example, he uses the number of jobs created every month instead of the overall unemployment rate normalized for labor force. Nate is comparing growth numbers for Obama and Reagan when reagan had a far smaller population.

            I know he reads the responses because he has responded to a number of my points in subsequent posts.

          • shiloh says:

            Bart

            FOX News 9/24 – 9/26 1092 LV 3.0 48 43 Obama +5
            http://www.foxnews.com/politics/interactive/2012/09/27/fox-news-poll-voters-want-change-president-to-stand-up-for-free-speech/

            Is Fox News skewing their poll in Obama’s favor?

          • Bart DePalma says:

            The pollster Fox News uses last had a Dem +9% bias. It was actually one of the worst of the lot. So much for fair and balanced.

          • shiloh says:

            Bart

            http://elections.huffingtonpost.com/pollster/party-identification

            Not that any evidence will ever change your mind …

          • Bart DePalma says:

            shiloh:

            That HP chart shows a Dem +8% average skew on the polls INCLUDING the relatively sane partisan ID spread in the Rasmussen, Gallup and this week’s Battleground polls. If those latter polls were excluded, the Democrat bias would be above 9%.

            The partisan distribution in the last election cycle in 2010 was 35% GOP, 35% Dem, 27% Indi with the rest refused.

            You see the problem?

          • shiloh says:

            Bart, 2010 was a low turn-out mid-term in which (29) million 2008 Obama voters did not vote. 2008 exit polls indicated 39% Dem ~ 32% Rep ~ 29% Ind. If anything, con hatred for Obama notwithstanding and Romney being such a god awful nominee, Rep 2012 turn-out may be even more depressed as cons are totally deflated w/their train wreck nominee.

            http://www.cnn.com/ELECTION/2008/results/polls/#USP00p1

            Stay tuned! :)

          • Bart DePalma says:

            2010 was almost identical to 2004.

            You are correct that the only way Obama could win is for conservatives to stay home. However, I am very active in conservative and Tea Party social networks and they are chomping at the bit to vote.

          • shiloh says:

            “2010 was almost identical to 2004.”

            ?!?

            2004 = 122 million votes ~ 2010 = 91 million voters

            2012 will be like 2004 as both Willard and Kerry are god awful opponents against an incumbent C-in-C. But Romney is a lot worse than Kerry. Both Bush And Obama were/are wartime incumbents and a wartime incumbent has never lost.

            btw, how clueless is Romney not to mention the U.S. military or Afghanistan in his acceptance speech!

            >

            Top Romney Official: Mitt’s A ‘Lousy Candidate’
            http://www.politico.com/news/stories/0912/81772.html

          • Bart DePalma says:

            2004 and 2010 had nearly identical partisan participation rates

          • shiloh says:

            “And what we couldn’t help but notice was that in this brief intermission, a distinct mood had settled in — that the competitive part of the election was over, and that the GOP challenger, Mitt Romney, was toast.

            And the Gallup daily tracking poll, which had suggested a narrowing only a week ago, had broken back in the president’s favor as well. One of the states we’d all but penciled into Romney’s column from jump, North Carolina, was seemingly wavering. Obama was mulling a play for Arizona.

            In addition, the quasi-history of recent elections strongly suggests that the end of September is a point of no return in presidential politics, where the leader tends to hold that lead until the end. And while Romney actually had a good week, relatively speaking — he didn’t stride headlong into hidden rakes, a la Sideshow Bob of “Simpsons” fame — the significance of his “47 percent” remarks from prior weeks continued to resonate, and resonate more deeply.

            “By a 51-45 percent margin in Ohio, voters believe Obama is better equipped to handle the economy than Romney. In Florida, the spread is 51-46 percent. And clear majorities of voters in both states -– 56 percent in Florida and 58 percent in Ohio -– say that the economy is either improving now because of Obama’s policies or will improve because of them.”

            http://www.huffingtonpost.com/2012/09/29/sense-of-an-ending-mitt-romney_n_1924413.html

            Bart, as always, keep hope alive!

  5. Shiloh says:

    According to Mitt, 47% of the country is mooching Democrats who will never vote for a Republican. That’s going to make it tough for the GOP to win elections.

    • actually what mitt said was that 47% of people dont pay federal income tax which is 100% true

      • Bart DePalma says:

        Mitt was referring to the 47% level of voter support for Obama and then reeled off a number of characteristics of his supporters, none of which apply to all of them, including paying no income taxes. Millions of GOP supporters in the bottom half of earners also do not pay income taxes and millions of Democrats are among the upper middle class and wealthy who pay nearly all of the general revenue taxes.

      • Shiloh says:

        No, he said that 47% of the nation will vote for Obama no matter what. Then he tried to conflate those people with the people who don’t pay federal income tax, which is a pretty blatant lie. He also said that he can’t make those 47% (who don’t pay FIT) take responsibility for themselves, which has got to be pretty insulting to the 47% not paying FIT who will vote for him. Fortunately for him, a lot of them probably don’t realize that he’s talking about them.

        In short, it was a disaster for Mitt.

  6. Martin says:

    Mr. DePalma, you seem like someone committed to the truth, and to understanding the reality of what’s going on in the polls. Let me, then, presume upon your good graces a little further and offer a few issues I see with your assessment:

    First, I don’t see any reason to believe the cell-exclusion bias is constant from 2010 to 2012. The number of cellphone-only households has been rising rapidly since 2010. One can debate what the real number is (and there is certainly debate on that, with estimates ranging from 25% to 40%, with around 30-33% being the most accepted), but no one argues that it hasn’t been climbing fast. Since there is ample evidence that the cellphone-only population skews liberal/democratic, the logical assumption should be that excluding cellphone-only households will have resulted in an increased anti-democrat bias since 2010, not a constant bias.

    Second, Datechguy’s analysis seems to have some problems. Cellphone-only housheolds were effectively nonexistent before 2008, and even in 2008 were a small percentage. So all the data before 2010 doesn’t speak to any current Rasmussen bias/accuracy well at all. As for the 2010 elections, Rasmussen’s numeric predictions in the actual races (margin of win/loss) were among the worst in the polling industry. Yes, he showed a net positive Republican identification, and yes, the Republicans picked up a lot of seats, but there is no reason to believe that the general US partisan identification numbers needed to be just like Rasmussen said they were in order for that outcome to occur. Pew’s partisan ID numbers for the general population were VERY different than Rasmussen’s, yet Pew called the numerical win margins in the races MORE accurately, not less. So it’s hard to read Datechguy’s post as a compelling defense of Rasmussen.

    Third, since Rasmussen is out-of-family with almost every other poll out there, one would need to offer a reason why the methodology of all the other polls are genuinely flawed/biased in order to bolster a claim that Rasmussen is really the accurate poll. I’ve looked for such a reason, and I can’t find it. The vast majority of them are NOT correcting to 2008 voter turnout models (despite what some have said), so that isn’t the problem. What they do correct for, reliably, is demographics. But those numbers are drawn from the census, and I believe Rasmussen uses the same source data for demographic corrections. What error, then, explains why all of the other polls are the biased? I have never found, nor seen, a compelling explanation on that.

    If we acknowledge that Rasmussen has a bias (although we don’t know how large exactly), and we don’t have a mechanism to explain why all of the other polls would be biased too, why should we not believe that the known Rasmussen bias is the reason that it is out-of-family from the other polls?

  7. Bart DePalma says:

    Martin:

    Thank you for your detailed post. Please allow me to respond.

    There is indeed a furious debate over whether landline, cell or various hybrid samples are representative of the electorate. Frankly, I have not heard a good answer from any pollster, so I continue to stay with the pollsters whose product reasonably approximates the final vote.

    Rasmussen is only one of a number of polls I personally credit. Battleground is excellent. Gallup is good once they get their likely voter screen up. I liked PPP in the past, but they are putting out nothing but double digit Democrat biased polling these days.

    As to partisan participation, please note that I did not use the Rasmussen registration numbers as the baseline in my back-of-the-envelope correction of last week’s polls. While both Rasmussen and Gallup have recorded a drop in Democrat registration since 2008, Rasmussen may by a point or so too good to be true. Instead, I conservatively used the historical Edison exit polling, which repeatedly undercounts conservatives. In that polling, 2008 was an outlier election and otherwise the percentages of participation simply do not vary that much.

    Finally, Pew is an Democrat outlier even among those liberal political forecasters like Nate Silver who credit the skewed media polling. I suspect that their findings will suddenly change in the last couple weeks before the election.

  8. Kernil says:

    Hello Bart,

    I assume you will follow-up after the election with an apology if you’re wrong and all of the polls (sans Rasmussen) are correct?

    • Bart DePalma says:

      What is up with you folks on the left with apologies as if disagreeing with you is a thought crime?

      In any case, my corrected presidential polling average does not even include Rasmussen.

      • Kernil says:

        I don’t know how to respond, because your response is a complete non-sequitur…. Just like your analysis of the polling data i suppose. i really couldn’t care less who you do or don’t agree with. Your analysis however, disagrees with all of the facts.

    • Pete Kent says:

      Funny but Ras has Obama ascendant of late, while it is in Gallup that the Presiden’t numbers have collapsed. If this keeps up the liberals are going to need a new polling whipping boy.

      @petekent01 on Twitter

      • Bart DePalma says:

        On the three day Rasmussen rolling poll, Obama apparently had a couple good days on Monday and Tuesday, probably related to the 47% attacks. Monday fell off the average today, bringing it back to Obama +1 and Tuesday will fall off tomorrow.

  9. Pete Kent says:

    I would like to add my 2 cents (more like 200) by posting my quantitative analysis of oversampling in the post-convention polls. I tried to post this in full to Nate Silver’s blog, but it was far too long . . . .

    Analysis of poll samples as compared with voter registration figures and exit poll data over the past few cycles indicates that Marist, Quinnipiac and Fox state polls are using very aggressive turnout models that are tilted towards the Democrats.

    For instance the Marist polls released on September 14 for Virginia and Ohio had significant Democrat oversamples on the order of 8 percent for VA and in excess of 10 percent for OH.

    The Quinnipiac polls that came out on September 19 had the same sorts of problems. In VA Democrats are oversampled by 14 points when you compare the poll sample with actual voter registration (VA GOP has net 3 point registration advantage, but poll shows Dem sample to be +11 over GOP). If you used the right sample, Romney wins the Old Dominion in a walk. Consider that this same poll has Kaine up by + 7, a clear outlier. Most other polls show an essentially tied race. The Quinnipiac Wisconsin poll was not as bad with only a 4 pt Dem oversample, but still large. Only in Colorado, where the race was shown as tied do they use a sample that has only a slight bias vs voter registration (D +1).

    The Fox polls from 9/20 were similarly challenged. In OH the sample had a pro-Democrat bias of 14% compared to voter registration numbers. The poll has Ds turning out by a net 3% more than they did in 2008. The problems were a little less severe in VA with a bias of D+6. The FL poll was perhaps the best of the 3 but still had Ds turning out by a net 2% more than in ’08.

    Yesterday’s Marist polls can be similarly decompiled and debunked. The IA poll oversamples Democrats based on registration and exit poll history by substantial margins essentially equal to the Obama lead. When you have 90% of Democrats voting for President Obama each extra percent of Democrats sampled adds about that much to the President’s lead. Exit poll data over the past 3 cycles does not justify the substantial Democrat lean in the poll. Republicans turned out in excess of Democrats by 2 points in 2004, while Democrat turnout exceeded Republicans by only 4 points in 2008 (an extraordinary year for Ds) and only 2 points in 2000.

    The Marist CO poll looks particularly bad – it oversamples Democrats based on registration and exit poll history by substantial margins in excess of Obama lead. The poll sample included a net excess of Ds over Rs of +2 (expands to +6 w leaners). But if you look at voter registration data for CO the GOP actually has the advantage by +3. Looking at exit poll data over the past 3 cycles Rs turned out in excess of Ds in each election, by 4 in 2008, by 9 in 2004 and by 3 in 2000.

    The Marist poll in WI looked pretty clean, showing a 5 pt lead for the President there – about what you might expect from a state he took by 14 points in ’08.

    The National polls are no better. CNN published a post convention poll that showed Romney had a huge advantage with Inds and then included only 4% in the sample! The latest Pew Poll was a real stinker. It had a net Democrat lean of +9 (+8 with leaners). Only by oversampling Democrats can you get the President lead to 8 percent over Mr. Romney.

    Some perspective: The electorate in 2008 was, in fact 7 percent more Democratic than Republican, but in 2000 the Democrat advantage was only 3 percent and in 2004 it was zero (same as in the 2010 Midterms that produced a GOP landslide). If you adjust sample to a more realistic electoral composition (no one is predicting a turnout model like 2008) then the race is either close to tied or Mr. Obama has a few points lead.

    @petekent01 on Twitter

    • Bart DePalma says:

      Great analysis. Feel free to leave your long posts here.

      Hugh Hewitt interviewed the Marist pollster on his radio show and the pollster refused to say how Marist consistently obtained a Democrat heavy sample without weighting the results.

  10. The Recurring — and Misleading — Focus on Party Identification
    By Frank Newport, PhD. – Editor-in-Chief of Gallup

    The discussion of the party identification composition of poll samples comes up in every presidential election with which I’ve been involved. Interested observers often opine that when a given poll shows that Candidate X is ahead, it cannot be correct because there is a higher percentage of voters who identify with Candidate X’s party in the sample than there should be, based on comparison to some previous standard.

    There are several reasons why this is a faulty approach to evaluating a poll’s results.

    Party identification is basically an attitudinal variable, not a stable population parameter. It is designed to vary. This is distinct from demographic variables such as age, gender, race/ethnicity, and education, which are, generally speaking, stable indicators measured by the U.S. Census Bureau. The only issues relating to demographic variables are measurement concerns — e.g., how the census, which creates the targets, measures ethnicity versus how individual pollsters measure it. But, generally speaking, these are fairly stable targets.

    Party identification is not measured by the U.S. Census Bureau, nor are there any other official state or national standards for what party identification “should be” in terms of the percent per party as it relates to the general population.

    Many people use the exit polls as a standard. But exit polls use a distinct question wording, a different methodology (in person interviews at the polling place as opposed to telephone interviews), a different environment (people are asked their party identification just after having voted, which could affect how they answer), and different sampling techniques to develop who it is that is asked the question. So party identification figures as measured by a specific poll aren’t easily compared to party identification as measured by an exit poll because of these and other potential issues.

    Party identification changes as political tides change. General shifts in the political environment can affect party identification just as they can affect presidential job approval and results of the “Who are you going to vote for?” question.

    Here is how Gallup asks party identification: “In politics, as of today, do you consider yourself a Republican, a Democrat, or an independent?”

    Note that this question does not ask, “What was your party identification in November 2008?” Nor does it ask, “Are you registered with one party or the other in your state?” Our question uses the words “as of today” and “consider.” It is designed to measure fluidity in political self-identification.

    We know that party identification moves over time — sometimes in very short periods of time, just like other political variables. Generally, if there is a political tide toward either of the two major parties, all questions we ask that are of a political nature will move in that direction. This includes the ballot, job approval, party identification, among others.

    So, it would not be surprising to find that if Barack Obama is enjoying a surge in popularity in any given state, that surge will show up on the ballot question, on his job approval measure, and on the measure of party identification. So, data showing that Obama is ahead on the ballot in a specific state poll and that Democrats have a higher-than-expected representation on the party identification question, are basically just reflecting two measures of the same underlying phenomenon.

    This doesn’t obviate the possibility that a sample is a “spurt” or a sample that happens to pick up higher than usual support for one candidate or the other for whatever reason. But, if it is a spurt, the cause is not “getting too many Democrats/Republicans in the sample.” It is instead a matter of “Getting too many people who, in response to all political questions, answer in a more Democratic or Republican” way.

    Basically, if an observer is concerned about a poll’s results, that observer should skip over the party identification question and just look at the ballot directly. In other words, cut to the chase. Don’t bother with party identification sample numbers. Look directly at the ballot.

    For example, we know that in Ohio:

    Obama won by 5 points in 2008
    Bush won by 2 points in 2004
    Bush won by 3 points in 2000

    Now if a given poll in Ohio in this election shows Obama with a 10-percentage-point lead, one should just ask, “How likely is it that Obama would be ahead by 10 points if he won by five points in 2008?” — forgetting party identification, which we assume is going to be higher for the Democratic Party if Obama is ahead, anyway. The discussion of the ballot in the context of previous ballots is, in fact, a reasonable discussion. It may be unlikely that Obama will double his margin in 2012 from what occurred in Ohio in 2008. Or maybe not. But the focus should be directly on the ballot, and discussions of reasons why it might be different than one expects should not involve an attempt to explain the results by focusing on changes in party identification — which is basically a tautological argument.

    In Florida:

    Obama won by 3 points in 2008
    Bush won by 5 points in 2004
    Bush won by [much] less than one point in 2000.

    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. The finding of differences in party identification is, instead, simply reflecting what one sees on the ballot.

    Essentially, it is much more direct to just focus on the trends and comparisons of the ballot question than it is to introduce an extraneous look at trends in party identification.

    I’ve been analyzing election surveys at Gallup since the 1992 presidential election, and I don’t personally put a great deal of stock in survey-to-survey variations in party identification. All of our weighting focus is on the effort to bring more solid demographic variables into alignment with census figures — including in recent years cell phone and landline phone use. We don’t find that party identification is stable enough to be of much use when it comes to comparing sample-to-sample variations, or sample to exit poll differences.

    • Bart DePalma says:

      We know that party identification moves over time — sometimes in very short periods of time, just like other political variables.

      Party registration does not move very quickly at all. That is the question Gallup (and Rasmussen) ask, but most other pollsters do not. Even the more subjective party self identification does not move as much as these media polls indicate. Most folks keep the same party identification their entire lives. Keep the question to party registration.

      “I’ve been analyzing election surveys at Gallup since the 1992 presidential election, and I don’t personally put a great deal of stock in survey-to-survey variations in party identification.”

      I wouldn’t either if the variation was a product of random chance and went in both directions equally. In fact, the variation only goes in one direction – toward overcounting Democrats and undercounting Republicans. There is no contra example.

      “All of our weighting focus is on the effort to bring more solid demographic variables into alignment with census figures — including in recent years cell phone and landline phone use.”

      THIS is why pollsters are overcounting Democrats and undercounting Republicans. Instead of determining the partisan breakdown of the pool of registered voters – the only folks who can legally vote, Gallup and most other pollsters apart from Rasmussen and Battleground are increasing the weight of Democrat-leaning minorities like African Americans and most especially Hispanics in their polling far beyond their actual share of registered voters by using the census shares of the population including well over 11 million immigrants who cannot legally vote and millions more who are not registered to vote.

      If media pollsters identified the partisan weighting in the only pools of respondents that matter – Americans registered to vote – and then weighted their respondents to coincide with how they are registered before applying a likely voter screen, these garbage polls might maintain some accuracy.

  11. Think we would forget about this after the election? You republicans live in an alternate reality. One in which you change the facts to fit your beliefs. You are a lightweight going up against statisticians that do this for a living. You should concede defeat, but you will not. So go ahead and try to spin this as to somehow you were right anyway.

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