Poll Insider Methodology

Pollinsider.com provides independent analysis of polling data regarding elections, most notably the 2012 Presidential Election. We do not have our own polls but instead analyze information gathered by other polling units, such as Public Policy Polling, NYT, CBS, FOX, Quinnipiac, and so on. Why do we do this? Most polling outfits do not analyze their results in a way that accurately reflects the population at large. PPP is both the most prolific pollster and the most useless polling outfit when it comes to base numbers. Nearly every poll heavily oversamples one party over another and gives misleading impressions. We correct that and lay out our methods for you to see and agree with.

Methodology: When enough information is given (PPP is great when it comes to providing crosstabs) I am able to re-calculate numbers based on reality. For 2012 Polling, I use data from the most recent election’s exit polls to properly assign the correct % of respondents from each party, re-run the numbers, and come up with a more realistic number to show where a race is at at any given point. For the most part, I use 2010 data which was an admittedly strong Republican year. It should be anticipated that this trend would continue through to the 2012 election cycle, just as 2006 and 2008 were similar and 2002 and 2004 were similar. A deeper discussion on the usage of 2010 for polling numbers can be found here. This will be a completely scientific, likely-voter turnout model and makes no assumptions. This is Level 1 analysis.

After recalculating to eliminate bias, in my level 2 analysis I assign “undecided” voters to the candidates to eliminate the Incumbent Skew. Incumbents usually have a much lower “undecided count” from within their own party since they know that the incumbent is going to be the nominee, and that they have no other challenger to consider. Challengers in a contested primary usually will see the undecided # being anywhere from 2-6 times higher as their are multiple candidates and respondents are more likely to answer undecided when their candidate is not in the head-to-head match-up being polled. This can often result in an artificial poll swing between 5-9% points in the incumbent’s favor. I recalculate the undecideds in equal fashion to the rest of the part vote. Finally, we assign the few undecided independent vote equally to both the Incumbent and Challenger, even though historically they swing to the challenger by 4:3. This calculation uses a scientific base and adds generally accepted assumptions about how undecided Democrats, Republicans, and Independents might vote.

Bias: I am a conservative. I work for no campaign. I use no advertising. I request no money. My candidate bias will be apparent. However, I carefully lay out all arguments objectively for others to analyze. My goal is not to create the illusion that a candidate is actually stronger (or weaker) than he or she actually is. My goal is to make sure cadidates are accurately analyzed and to retort misinformation given out by polling agencies who refuse to analyze their own data effectively.

  1. Harry Giarratana
    March 20, 2011 at 9:10 pm

    Thank-you for your insightful piece on Palin 3/20

  2. May 27, 2011 at 7:57 am

    I have just discovered you site and have posted it on my website. I’ve known all along that the polls were rigged, but until finding your site, had no idea where to find truthful interpretations. Thank you very much.

  3. June 6, 2011 at 11:29 pm

    The amazing thing about numbers is that they can be manipulated to for whatever conclusion one seeks.

    However, when numbers are left alone and reported exactly as intended the answers they give are as accurate as can be!

    What I have seen, you do an excellent analysis!

  1. June 3, 2011 at 3:24 pm

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 )

Google+ photo

You are commenting using your Google+ 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 )


Connecting to %s

%d bloggers like this: