Today we introduce a new feature to your favorite site on politics: The Political Derby Composite Poll. This tool takes the most recent major polls and weights their results based on poll age, sample type, and margin of error. Unlike other poll compilations, this is not a simple average that one skewed poll can unduly influence, rather it is a proprietary mathematical formula that derives the percentage of voters favoring each candidate, recalculated on a daily basis as polls age.

The polls that make up the PD Composite are well-known reputable national polls who sample “likely primary voters” and “registered voters” including, but not limited to: NBC News/Wall Street Journal, Pew Research, Quinnipiac University, ABC News/Washington Post, and Rasmussen Reports.

The PD Composite will be updated in a regular post that will run at least weekly, or more often if new polling data changes results significantly. This feature will also always be available at the “Composite Poll” link at the top of the page. Going forward, the Composite Poll will also be a significant factor in determining the Power Rankings.

Now that explanations are out of the way, let’s get to the Political Derby Composite Poll standings:

Note: Click the graph to enlarge.

For simplicity’s sake, we are only showing the Composite results back to September 1, when Rick Perry was surging. From there, you can see his decline. You can also see Herman Cain’s huge increase in popularity that coincided with Perry’s late September plummet, until they converged before continuing in opposite directions. Mitt Romney has been consistent, if not seeing a slight increase in popularity, as he runs a general election campaign throughout the primary. But the trend line that may be most worth watching is that of Newt Gingrich. Newt is the only person out of the bottom of the pack since Cain that has managed to distinguish himself through the debates and pull out of a flat-line trend. We could very well see him surpassing Rick Perry before the end of the year, becoming a top three contender.

Going forward, we hope you will enjoy this additional data-driven approach to evaluating the horse race here on


  • Coach Outlet Store Online

    This is a very good idea! Just want to say thank you for the information, you have to share Coach Outlet Store Online
    . Just continue to write such a position. I will be your faithful reader. Thank you again.

  • Coach Factory Online

    As Coach Factory Onlinepoints out, the Newt fall can come anytime. I still like Mitt as the odds on favorite due to his dicipline and organization

  • Edgar Harris

    Good job Scott.

  • Brian H

    I like the idea. I only hope it does not become the driving factor behind the power rankings. What I like about the power-rankings is that it includes the human element and takes into account the many characteristics that polls can’t determine.

    • Troy La Mana

      I agree with you but I think this data needs to be weighed heavily in the next rankings.

    • Scott A. Robinson

      We are no BCS, but the Power Rankings certainly does have a human element involved. The Composite Poll will be used as one of the criteria in establishing the Power Rankings.

  • T Baker

    OK, so what I’m wondering is… if this takes into account the “margin of error” of each poll, isn’t there a separate standard error inherent in combining the multiple poll results – a sample of samples – which would give you an additional margin of error?

    • Scott A. Robinson

      Ah, this would be the case if we were merely combining polls. However each poll is part of a weighted calculation. The margin of error portion of the calculation is actually the least significant, as each poll is statically significant. Where the margin of error differences typically come is from sample size, and they are not usually more than a two point spread.

      • Rick B.

        Don’t try to frighten us with your sorcerer’s ways!

        • Scott A. Robinson

          But we haven’t even talked about confidence level yet!

          Fortunately, as far as I have seen, all of these polls use a 95% confidence level, though you have to really dig into the details to find it because confidence levels are usually not reported.

  • Alaina

    So whats the difference between this and what Real Clear Politics reports?

    • Whodat

      This one is here and it is very good. And, I can leave a comment, for whatever that is worth, while I am in the neighborhood.

    • Scott A. Robinson

      There are some distinct differences between the Political Derby Composite poll and the RCP Average. RCP is a true average of polls from the past 7 days. The Composite is a weighted average giving more value to the most recent polls, more value to “Likely Voters” than “Registered Voters”, and slightly more value to polls with small margins of error. Also, we are not using any polls that poll simply “Adults”, as these are less likely to be accurate than those using “Likely Voters” than “Registered Voters”.

      We are using better data sources and giving more weight to polls that have more meaningful components or more accuracy.

  • T Baker

    Scott – where were you when the BCS was getting started?

    • Scott A. Robinson

      I’m still available if the BCS were interested. However, they’ve got their system tuned to do exactly what they want it to do, which is to continue feeding the schools from the politically well-connected conferences while hosing the Boise States of the world.

      • Gary Russell

        Warning: Totally off-topic comment ahead!
        Roll Tide!

  • Alaina Segovia

    Nice job Scott. It will be interesting to see how the Cain story will impact the polls over the next week or so.

  • Rob

    Pretty cool. You should do the same for Iowa, NH, SC and Nevada.

    • Scott A. Robinson

      If only it were that easy….

      But I might look into it.

  • Stephen Fountain, Editor

    Following that thought, Cain attratction out of no where is well represented. But could it be a sugar high?

    • dw

      Not so fast, Stephen. Cain has been in the race a lot longer than Sept 1, and was polling higher pre-Perry. I’d like to see the numbers going back to much earlier in the race.

      • Scott A. Robinson

        Other than a couple of blips in June, Cain has never polled in double digits.

        • dw

          I seem to recall otherwise. Do you have the composite info going back to when he entered the race?

          • Scott A. Robinson

            Although I haven’t modeled it, the data is clear. Cain’s best showings before late September were 10% in July in Public Policy Center poll, 17% in a June PPP poll, 12% in a June NBC/WSJ poll, and 10% in a June Rasmussen poll. He has never done better than that. The debate where he really pushed 9-9-9 and was able to connect personally with millions of Americans as a cancer survivor was clearly his turning point.

  • Rick B.

    Nice work!
    I’m no medical professional, but looking at this chart it appears that someone needs to hit Huntsman with a defibrillator.

    • Stephen

      He was never alive to begin with…

  • Gary Russell

    Good job, Scott.
    The evidence is overwhelming that Huntsman, Santorum, and Bachmann should give it up.
    At the least, their inability to hit 5% should keep them out of future debates. That would help to cut out a lot of the meaningless “noise”.

    • Alaina Segovia

      Totally agree… when are the debate moderators going to institute that rule?

      • Scott A. Robinson

        The only rule I know of is in the CNN debate, you had to have an average of 2% in several national polls. Huntsman was at risk of not making the 2%, so before the polls had a say in it, he determined he was going to “boycott” the debate for some odd reason I don’t even remember.

        • Alaina Segovia

          The one DeMint did in early September (that wasn’t televised… at least that I could find) required 5% or above, but after a certain point in the race (certainly at least by now) they should require candidates to have 5% or more.