• comment on a post Polling Project: First Numbers over 8 years ago

    My compliments on your polling efforts, I look forward to your results.  I posted this comment and wanted to offer it to you and your readers as interest.  Cheers...

    From Miguel Lecuona - Texas Pundit

    Polling vs Elections: Why the Polls Fall Short

    Most polls do not translate very well to presidential elections, in my opinion. Even well-intended polls by partisans, with sound methodology, leave lots of guesswork when attempting to project results to actual elections.

    Let's agree that this poll and the methodology are accurate for what they are asking and to whom they ask the questions. Let's also assume that there is no media bias or partisan polling bias in the results. I know -- stay with me anyway!

    The issue is projecting the results of this poll (or any poll) into a reliable election prediction. Todays do not speak to voter turnout, or to the changing voter demographics that will exist nearly 3 years from now.

    Also, I would be concerned that a sample of 1004 voters is not very translatable to Electoral College results.

    For example, Dems are heavily concentrated in major Northeast and California Urban centers, often carrying those cities by as much as 75% and creating big state-specific landslides. The poll says that, nationally 33% of all people are Democrats. If that's true, then this base is not widely or evenly distributed across the country, making it less impactful to an Electoral College majority. As soon as the Dems achieve more than 51% in a state, the rest of the votes are "wasted", so excessive margins indicate the base is highly concentrated. OK for polls, not good for Elec. College.

    Looking closer at the samples, they are deemed representative of the party affiliations... if that is the case, then the following holds true:

    Poll 1004 Voters
    Dems 33%
    Reps 29%
    Indep 37%

    2004 Election results: Bush 51%, Kerry 48%.

    Assuming both parties held their bases, this implies that Indeps voted 22 for Reps vs 15 for Dems.

    Are they really "independent" if they vote 3-2 for Bush? If they were Independent, one might expect a closer split. But if Independents continue to vote Republican, the distinction is not very meaningful.

    I will look forward to seeing the rest of the questions, and their answers. But the best polls are the actual election results.

    Miguel Lecuona - Texas Pundit


Advertise Blogads