Skip to main content

View Diary: New CO ARG Poll- Romney +1 48-47/ Claims 42% of Hispanic voters are for Romney! On what planet? (14 comments)

Comment Preferences

  •  there is ALWAYS something wrong in the crosstabs (2+ / 0-)
    Recommended by:
    JFactor, TrueBlueDem

    Unless you take an absolutely massive poll with thousands and thousands of participants,  there will be enough ways to slice and dice the electorate that after-the-fact you can find a bucket or two with results that look off.

    This is a reasonable looking lean-R poll that suggests what we could see in a worst case scenario - but the trendline is good and the election is next week.  (And no Sandy in Colorado!)

    •  You seem not to grasp (1+ / 0-)
      Recommended by:
      immigradvocate

      that they have assumed a 42% Hispanic R share to extrapolate the topline! Even if there are more evangelical, conservative Hispanics there than nationally (which I highly doubt), that wouldn't even get above 30%!

      •  okay, I'm not getting into the weeds on this one (1+ / 0-)
        Recommended by:
        antooo

        Obviously, if everything else in this survey came true on election day, except that the Latino vote turned out big and voted for Obama in the margins that you predict, then yes, Obama would take Colorado.

        But.... there are almost certainly other crosstabs that are  "off" and could be used by a right winger to argue that Romney really has a bigger lead.

        My point isn't to debate who has the right take on the crosstabs and whose adjustments are correct.

        My point is that obsessing over "correcting" weird looking crosstabs is the first step to the madness of poll-unskewing. Statistically, there will almost always be sub-populations that look off in all but the largest polls. Combing over these is not productive.

        your larger point remains:  A big Latino turnout and Obama wins Colorado.

Subscribe or Donate to support Daily Kos.

Click here for the mobile view of the site