Skip to main content

View Diary: So, about that CNN Snap Poll (290 comments)

Comment Preferences

  •  But (5+ / 0-)

    100% of debate watchers were not from the south, but the respondents to this poll are all from the south.

    •  Again, just so people aren't confused, (3+ / 0-)

      The reason the other regions say N/A in the columns of the internals is only because the margin of error for those subsets of the poll is too high to be meaningful.  And actually the South is now the most populous region of the country, although it does look like it was overpolled here.

      •  But (5+ / 0-)

        the entire under fifty crowd, as a block, is also outside the margin of error? Same withe the non-white block? I realize we have a lot of seniors in the country, but seriously?

        •  Only 430 people in this poll, as opposed to 1,000+ (5+ / 0-)

          in normal national polls, so the margin of error is high to begin with, and thus it makes even releasing internals mostly pointless.

          Although if they had polled an accurate number of non-white people, the margin of error would've been small enough to report the numbers.  It looks like they're only including internals of subsets with a MoE below 10%, which I believe is the common number among pollsters.

          •  Agreed. (2+ / 0-)
            Recommended by:
            typo ink, NotGeorgeWill

            "When fascism comes to America it will be wrapped in the flag and carrying a cross." - Sinclair Lewis

            by Bob Duck on Thu Oct 04, 2012 at 12:22:30 AM PDT

            [ Parent ]

          •  so then, (1+ / 0-)
            Recommended by:
            wu ming

            Less than 10% of their sample was in any age group under 50?
            Less than 10% of their sample was Hispanic?
            Less than 10% of their sample was from the West, Northeast, etc.?
            Seems hard to believe that this was random. There seems to be selection bias.

            "We have always known that heedless self-interest was bad morals; we know now that it is bad economics" - F.D.R.

            by biscobosco on Thu Oct 04, 2012 at 12:59:03 AM PDT

            [ Parent ]

            •  You're misinterpreting margin of error (2+ / 0-)
              Recommended by:
              ItsSimpleSimon, tiredntexas

              The poll's overall sample size is 430 people, which has a margin of error of 4.7%.  As you poll fewer and fewer people, the sample size gets bigger and bigger.

              In this particular sample, any group that contains less than approximately 100 people, or 23% of the sample (I'm doing a reverse calculation here), will have a margin of error of 10%+.  

              So again, the poll has issues, but it's not as bad as it seems at first.

              •  huh how does the sample size get bigger (1+ / 0-)
                Recommended by:
                typo ink

                as you poll fewer people?

                Other than that, what you say makes sense and thanks for explaining. I thought it was 600, not 430.

                In that case, it would indeed be interesting to see the true data, the actual demographics.

                Also of interest is that although the US West is about as populous as the South? we know that less than 23% of the sample was from the West.  Which certainly means either that the sample was skewed, or the viewership was skewed.

                But I suppose the main takeaways are that the poll is not large enough to have a very good MOE on any of the results, and secondly we don't know if it oversampled Romney voters, or whether in general more Romney voters watched the debates.

                "We have always known that heedless self-interest was bad morals; we know now that it is bad economics" - F.D.R.

                by biscobosco on Thu Oct 04, 2012 at 01:43:59 AM PDT

                [ Parent ]

                •  Ah, typo! Meant to say margin of error gets bigger (0+ / 0-)

                  Now you see how I get my name.

                  The South is the most populous region of the US.  That's why I didn't want to criticize that aspect of the poll too much.

                  Here's the regional breakdown of the electorate from the '08 exit poll:  Northeast 21%, Midwest 24%, South 32%, West 23%.

                  •  I think the west is now more populous than that (0+ / 0-)

                    But probably still only around 25%  or so..

                     the basic point you are making is that there would be no crosstabs for any demographic group which was less than 23% of the sample..

                    Since you know a fair amount about polling.. I'm wondering whether you think 430 people is a reasonable enough size sample.

                    (If the viewing audience was 56 million similar to last debates)

                    "We have always known that heedless self-interest was bad morals; we know now that it is bad economics" - F.D.R.

                    by biscobosco on Thu Oct 04, 2012 at 02:16:23 AM PDT

                    [ Parent ]

                    •  A sample size of 430 people creates a MoE of 4.7% (0+ / 0-)

                      which is pretty high, but acceptable.  But as we've discussed, it makes the crosstabs mostly useless.  I'm guessing CNN polled such a small sample so they could complete the snap poll and breathlessly announce the results as quickly as possible.

                      Normally pollsters use a sample size of at least 1,000 or so for a national poll.  You usually only see samples of 400-500 in state polls.

                      •  ok, so that MOE (2+ / 0-)
                        Recommended by:
                        wishingwell, tiredntexas

                        Makes any change in approval ratings essentially meaningless.
                        I still believe there was some kind of selection bias in this poll, but as mentioned elsewhere, Nielsen would tell us if that is true.  

                        Do you know if you need a subscription to get that data?

                        "We have always known that heedless self-interest was bad morals; we know now that it is bad economics" - F.D.R.

                        by biscobosco on Thu Oct 04, 2012 at 02:34:49 AM PDT

                        [ Parent ]

                      •  note it was a subsample from a previous poll (0+ / 0-)

                        Offhand I would guess they attempted to reinterview everyone from that poll who was qualified and willing.

                        We should be grateful to them for that, since it gives us an interesting result: the candidate favorables hardly budged.

                        Election protection: there's an app for that!
                        Better Know Your Voting System with the Verifier!

                        by HudsonValleyMark on Thu Oct 04, 2012 at 03:47:18 AM PDT

                        [ Parent ]

      •  hard to buy that (2+ / 0-)
        Recommended by:
        wishingwell, elmo

        Geography is not random.  You can get a geographically diverse sample.  The area codes are very helpful that way.

      •  Exactly. (1+ / 0-)
        Recommended by:
        Bob Duck


      •  South only represents about 1/3 of the population (1+ / 0-)
        Recommended by:

        So if 50 million people watched, (about 6 million less than in 2008) and 90 % were over 50 and registered voters in the south, I suspect we are reaching a state of impossibility.  

        90% of 50 = 45 million  

        Lets say 110 million in the south, and charitably, 35% of those are over 50. Of course we have to leave out black people, lets say they are 20% of southerners.  
        so 110 - 20% = 110 - 22 = 88 million whites * 35% = approx 31 million white southerners over 50.

        So if 100% of  white southerners over 50 were watching the debate, it still would not be enough. They would be short  by 15 million.

        This is absurd.

        "We have always known that heedless self-interest was bad morals; we know now that it is bad economics" - F.D.R.

        by biscobosco on Thu Oct 04, 2012 at 12:37:45 AM PDT

        [ Parent ]

      •  w/430, N/A= so few, might as well be 0. (1+ / 0-)
        Recommended by:

        Srsly, stop trying to defend this bad sample.  

        Does it mean BO won the debate. No.  But it also doesn't mean the narrative CNN is obviously trying to set against BO is true either.

        •  Actually, with only 430, a subgroup of 100 is N/A. (2+ / 0-)
          Recommended by:
          ItsSimpleSimon, tiredntexas

          That's my point.  With the overall sample size so small, a subgroup that constitutes 23% of the sample will show up as N/A.  So it's possible that the West, Northeast, and West all constituted about 22-23% of the poll, and that would actually be reasonably accurate.

          I agree that the sample is bad overall, and I'm not defending it.  I'm just trying to help people get it right so they don't attack the poll on an incorrect basis.  Attack the poll because it's too old or too white, but don't attack it because it "only included Southerners".

          •  Sure, but if 3/4s was southerners and they over-fa (0+ / 0-)

            vor Thug/1% by, say, 30% compared to all other sectors of the country, that would skew overall results 1%'s way by a (back of the envelope) relative 10% of so.  Add additional 1%/Thug bias from oversampling other friendly cohorts and much of the difference btwee this and the CBS poll disappears.

            Sure, that's not good, but its not the crushing blowout CNN is selling.

      •  Oh, come on (3+ / 0-)
        Recommended by:
        wishingwell, elmo, tiredntexas

        They sought out this demographic group.  It'd be OK (weird but OK) so long as they consistently labeled it as such, including in headlines.  They did not.

        Pro-Occupy Democratic Candidate for California State Senate, District 29 & Occupy OC Civic Liaison.

        "I love this goddamn country, and we're going to take it back." -- Saul Alinsky

        by Seneca Doane on Thu Oct 04, 2012 at 03:12:15 AM PDT

        [ Parent ]

Subscribe or Donate to support Daily Kos.

Click here for the mobile view of the site