December 17, 2007

What’s with the polls?

Of the last six national polls of the GOP field,

  • CNN shows Rudy with a 2% lead
  • CBS/New York Times shows Rudy with a 1% lead
  • ABC/Washington Post shows Rudy with a 6% lead
  • AP-Ipsos shows Rudy with an 8% lead
  • ARG shows a tie between Rudy and Huckabee
  • Rasmussen is the only poll that actually shows a lead for Huckabee…by a brazen 6 percentage points over Rudy.
  • The freshest poll, a USA Today/Gallup poll taken over the weekend, shows Rudy with an 11% lead over Huckabee.

All the polls claim a margin of error of around plus-or-minus 4 or 5%, but in looking at Rasmussen vs. USAT/Gallup, there’s a 16% discrepancy between their polling results. So, someone’s claim of a minuscule margin of error is bogus, and I’m suggesting it’s Rasmussen, since his is the only poll that shows a clear lead for Huckabee.

Ras is pulling the very same thing he pulled with Fred Thompson this summer and early fall, when his polls were the only ones showing Thompson leading the GOP field. Of course, the media loves Ras’s polls, but they are likely less naively falling for his numbers than they are picking and choosing his numbers because they create the kind of drama and confusion in the GOP race that the mainstream news media craves. And the others who love Rasmussen’s numbers are Rudy’s detractors, who have (once again, like they did when Ras was showing Thompson with a fictitious lead) pounding the drums over the demise of Giuliani’s campaign.

As we head into Christmas week, and then the week before New Years, most people won’t be paying attention to the efforts of some to game the system with questionable polls. But to Rudy’s detractors, beware of placing your bets on suspicious polling results. If Rudy makes a contest of it in New Hampshire, and South Carolina, and goes on to win in Nevada and Florida by convincing margins, his results will have the appearance of a “miracle resurgence.”

by @ 6:11 pm. Filed under Poll Watch
Trackback URL for this post:
http://race42012.com/2007/12/17/whats-with-the-polls/trackback/

25 Responses to “What’s with the polls?”

  1. ACT Blog Says:

    Rasmussen polls likely Republican voters, using a special screening system to get more accurate results. Since likely voters are more Conservative in the GOP, that favors candidate who do much better among Conservatives, like Huck.

    Gallip only polls adults, who are less likely to be paying attention, and more likely to be moderate. That gives the advantage to better known, more moderate candidates like Rudy.

  2. Matthew E. Miller Says:

    I think Rasmussen clearly polls more so-cons and strong conservatives in his samplings. This is why Romney’s numbers look similar across polls (he’s pulling support from moderates and conservatives equally, so it doesn’t matter which a particular emphasizes). It’s also why Huck and Rudy look so very different between Ras and other polls (right now, Huck’s considered the most conservative candidate in the field). I don’t know that I’d say Rasmussen is playing any “game”. He may well have good reason to believe he should screen more tightly for conservatives. I don’t know to what extent other polling companies have factored in the dissatisfaction with the GOP among independents and moderates; if it hasn’t played a large roll in their screening process, then Ras may well be the only guy who’s getting it right.

  3. Big S Says:

    I still maintain that Ras’s tight screen for “likely voters” is being contaminated by the weird 2006 midterm election, where Republicans in a lot of northeastern and moderate districts were wiped out in favor of Democrats, with only the most culturally conservative voters going (R). A lot of “likely voter” screens place an emphasis on past voting patterns to determine who is committed enough to vote in a primary election, and midterm elections, where turnout is depressed anyway, are a pretty good indicator of that. If anybody knows the details of Ras’s (or anybody else’s) questionnaire to determine likelihood of participation, speak up; that would help to clear up the discrepancy.

  4. Big S Says:

    #1

    Gallup generaly polls adults and then splits their sample into Republicans/leaners and Democrats/leaners (and sometimes likely voters). Rasmussen disingenuously claims that his polling firm is the only one that screens for likely voters, which makes me question his methodology.

  5. murphy Says:

    This is the effect of free press which has been so often dismissed by Rudy supporters. Huckabee gets lots of free press, so it’s no surprise when he grabs some of Rudy’s soft support.

    A month from now, when Rudy’s running 0-3 in early primary states, he’ll long to be back within the margin of error with Huckabee.

  6. davew Says:

    murphy,

    long to be back within MOE of huck??

    HAHHAHAHAHAHAHAHA….you’re a funny guy! Huck is ALREADY falling…i hope he enjoyed his two weeks!!

  7. Ajay Says:

    Greg – do you know statistics? Your claim that this is just a “margin-of-error” issue is totally false. There are 2 issues here. First, the margin of error as you correctly pointed out. And the second, is BIAS. Bias is basically an inherent flaw in the polling mechanism (i.e. systemically [i.e. always] polling too many moderates and not enough conservatives, or polling too many young people, too many evangelicals, etc.).

    A poll can have no margin of error but it might be biased. Here’s an example, I poll every adult in America and publish the results. My “poll” will have no margin of error but it sure as heck will be biased.

  8. Peter Says:

    Our pro-life President skips the annual march for life for the seventh year in a row.

    Imagine if he did that to the VFW and the NAACP?

    http://catholicsforhuckabee.blogspot.com

  9. Big S Says:

    Matthew E. Miller,

    If it was as simple as Rasmussen identifying dissatisfaction with the GOP, then Gallup would pick up a similar thing in their polls of Republicans and Republican leaners. The likelihood is that the difference between their polls is in Rasmussen’s idea of who is likely to vote among the already identified Republicans and leaners.

  10. murphy Says:

    davew,

    No way does Rudy hold his current national polling support after losing IA/NH/MI. January’s going to be rough for your guy.

  11. Gary Matthew Miller Says:

    I wish like h*ll that Rasmussen was wrong but he has, by far, the best track record of any major poll over the last few cycles.

  12. Big S Says:

    11

    It’s a good track record, but not *BY FAR* the best. In fact, there are a lot of other firms that are about as good, and a few that are significantly less trustworthy. The biggest difference between Ras and the others is that Scott Rasmussen is a shameless self-promoter.

  13. Matthew E. Miller Says:

    Big S,

    Oh agreed. I meant that assuming someone’s a Republican leaner, we still don’t necessarily know how likely they are to actually stop leaning and fall into the Republican column. You’re right that the type of screening is the key factor. For instance, if Ras asked all Republican leaners how conservative they were, on a scale from 1-10, and then weighted the sample extraordinarily to favor 8-10 or something of that nature, we’d certainly expect the sampling to favor the more conservative candidates. The only question is, would Rasmussen be “right” in doing something of this nature. For instance, various polls show that Huckabee is considered the most conservative Republican among the general population, and is supported by very conservative Republicans. But, I think if we were to a poll of Huckabee’s “firm” support (bloggers, endorsers, volunteers, etc) I think we’d find that they’re considerably less conservative. So in Huckabee’s case, it’d probably be reasonable to conclude that alot of the very conservative people supporting him in polls, are less likely then his moderate supporters, to actually get down to the polling station and vote.

  14. Argamenon Says:

    Look no further than the sample. Gallup polled 399 adults. You can toss that whole poll into the trash can.

  15. Big S Says:

    14

    Dude, it’s of “Republicans and Republican leaners”. The post on the front page here is wrong.

  16. econ grad stud Says:

    All recent national polls except GALLUP show an increase in support for Huckabee and a decrease in support for Rudy.

    In that sense GALLUP may be an outlier.

    I also don’t think you understand the idea of margin of error. It’s not that you’re within 4% of a candidate’s true support. It’s that 95% of the time you ought to be ‘off’ no less than 4%. 5% of the time you’ll be off. Sometimes you’ll be ‘off’ by a lot.

    In the case of GALLUP we have a poll that has a porous filter. Rasmussen has a tight one. In primaries typically only 1/6 to 1/5 of Republican voters actually participate. Unless the turnout is radically higher Rasmussen ought to be more accurate than the others if it has correlated it’s questions with likelihood of voting.

  17. Big S Says:

    13

    The problem with using something like self-identified degree of conservatism among poll respondents is that the term means very different things to different people. That’s why it’s useful to ask voters if they voted in the last election; it takes away some of that subjectivity in terms. However, it biases the weighting in polls towards the turnout of previous elections, and can cause a lot of variation when there are big swings in voting demographics. I haven’t been able to find Rasmussen’s “likely voter” questionnaire published on the web, but I’d bet they ask respondents about their recent voting history.

  18. Shawn Says:

    GO RUDY!!!!

    A Vote For Huck is a Vote for the Beast!

  19. Dominic Harr Says:

    Well on 1 Oct Raz had Fred leading the pack.

    So clearly, his polling method is *not* predictive of people’s votes.

    Kinda hard to ignore that . . .

  20. Tano Says:

    “So clearly, his polling method is *not* predictive of people’s votes.”

    NO poll is EVER predictive of actual voting. Polls, at best, are snapshots of what the people think at any one moment.

    Showing Fred in the lead at one point may or may not have been accurate. The fact that everyone knows that as of now, Fred is not in first, doesnt address the question of whether the population felt that way in the past.

  21. John Galt Says:

    its is called methodology DaveG. EAch pollster has differnet methodologies. Rasmussen as it is no secret screenf for ‘likely’ voters. other aks questions in different ways…. etc etc. that is why an average gives you a somewhat decent idea of a general consensus but by no means an accurate pinpoint assesment.

  22. Palin for VP! Says:

    Why are we debating whether Gallup or Rassmussen are correct? Isn’t it possible that both are wrong? After all, they are the two extremes of the spectrum.

    I hate to say this, but If I had to pick from the list at the moment, I would go with the news networks (ABC/CBS/CNN) – Rudy with a slight lead.

  23. JMC Says:

    Could the Gallup poll be a result of Rudy’s speech over the weekend? I can’t find when the poll was actually taken – whether over a period of days or what. Regardless, I’m glad to see it because imo Rudy is the best bet all the way around.

  24. IowaFan Says:

    Rudy is indeed the best bet all the way around.

    The Hucklebee surge is fleeting – it may last through Iowa – but he’ll soon crumble under the national media spotlight, with or without DNC’s help.

  25. TennJoe Says:

    Gallup could be finally picking up the after effects of all the revelations and critisims against Huck in the past two weeks.

    Let’s see what the next round of National polls from the other sources show to see if Gallup is the first to catch the bloom coming of Huck’s rose.

State of the Race


Obama Approval


Support R4'12

Meta

Recent Posts

Buy This Book

Categories

Archives

Search

Blogroll

Site Syndication

Main