Skip to main content

Andrew Gelman over at Fivethirtyeight.com purports to show how whites at various income levels voted in 2008at the presidential level, which is an interesting enough concept. However, Gelman inexplicably avoids using exit poll data to make his charts. Rather, he bizarrely decided to make broad assumptions about race in this country (a touchy subject under the best of circumstances) by using pre-election polling data from Pew as the basis for his regression analysis.

To reiterate, since I could hardly believe it myself -- Gelman based an entire study of the white vote in 2008 based not on, you know, election results, but on pre-election national telephone polling.

Originally I theorized that Gelman relied on the Pew data because it broke down the income into granular levels (under $20K, $20-40K, $40-75K, $75-150K, over $150K), and that it provided racial breakdowns for those categories. But looking closer, I found that 1) Pew didn't break down those categories by race -- Gelman filled in the blanks with statistical sleight of hand, and 2) counted eight income categories in the exit polling, so even more granularity than Pew provided. What's more, the exit polling helpfully provided race and income data for two categories -- whites (and non-whites) making over and under $50K.

In other words, there was a great deal more data, and BETTER data, that Gelman could've used to tease out the maps he created. So I decided to compare his results to the actual exit polling. The first state I checked was Colorado, knowing from memory that Obama easily won the white vote in the state despite Gelman claiming otherwise. My suspicions were immediately proven right: While Gelman claims only the under-$20K white demo went for Obama, the results were far different. Per the exit poll -- real voters -- Obama won all whites: 54-45 percent for those making under $50K, and 51-47% for those making over $50K.

So having one flubbed state under my belt, I decided to investigate further. Here are Gelman's maps:

Now remember, the exit polls don't offer race and income breakdowns at the level of income granularity Gelman worked with, but the map below shows, per the exit polls, how white voters making more than $50K per year voted.

What's different? Per exit polls, Obama handily won Colorado whites as already mentioned above. But in addition, he easily won them in New Hampshire as well, 54-45, despite Gelman's claim that Obama lost the $40-75K cohort. And given that New Hampshire is 94% white, and that Obama won all Granite State voters making $75-100K by a hefty 58-41 margin, I think it's safe to assume that there's little chance Gelman is correct.

On the other hand, check it out -- Michigan is red. Yup, McCain won whites making over $50K by a slight 50-48 margin. Are there enough Michigan voters making more than $150K a year to offset a supposed Obama victory among those making $40-150K? I bet a good analysis of the exit poll data along with your standard regression analysis could probably yield some interesting answers. That's what Gelman should've looked at, not pre-election telephone poll data.

I should note that Gelman seems to have nailed New Jersey, Maryland, Pennsylvania, and a few other of the more surprising results, which suggests his work had merit as a predictive effort, but the various misses make it unacceptable as a tool to explain election results. Telephone polling doesn't offer a solid foundation for making claims about election results.

Now let's look at numbers for poorer whites. First we'll look at Gelman's maps:

Below is the map of the under-$50K white vote based on actual exit polling.

Gelman has Obama winning Nevada in that (rough) income group, but the exit polls peg it an exact 48-48 tie. He has Pennsylvania Blue, when McCain won the under $50K white vote 50-49. Gelman has North Dakota red, which surprisingly was a 49-49 tie. And he has New Mexico blue, while the exit polls say McCain won 49-48. All of these are tight enough that they could tip one way or another based on the exit polling's margin of error, which is supposedly around 1 percent.

In Ohio, Obama won under $50K whites 51-47, and in Indiana, 50-49. Gelman's maps split these states, with Obama winning the poorest voters, and McCain the $20-40K group. We don't know if Gelman is right, but it's certainly possible. It would be better if we didn't have to guess.  

He has Connecticut red for the under $20K white vote, while Obama won the under-$50K white crowd by a solid 59-36. Looking at the income and race exit poll data in that state, it seems improbable that Obama would've lost the under $20K crowd.  His projections for Montana (where Obama won the under-$20K demo, but lost those making $20-40K) could theoretically be correct, but probably not. From the exit polls, Obama won the under $50K white demo 51-45. The state is 90 percent white, and looking at the overall income numbers, Obama won the $15-30K crowd 58-38, and the $30-50K group 51-45. With those results, it's hard to see how Obama loses the $20-40K white crowd.

New Hampshire is solidly Blue unlike Gelman's maps, 58-40 -- one of the most obvious misses in Gelman's analysis.

Look, I'll posit something up front -- I suck at math. Half the shit above, while at the "simple math" level, could be wrong. What I have a hard time understanding is why, given the wealth of exit poll data, someone would claim to make an analysis of the 2008 vote based on a pre-election telephone poll. I can't begin to fathom it.

Had Gelman made the same analysis based on this exit poll data, then all the power to him. Nate himself referred to exit poll data when he wanted to compare demo results. And yes, while exit polls have their own margin of errors and sample composition problems, they sure as heck beat anything done over the telephone. If nothing else, sample sizes of about 8,000 per state of confirmed (as opposed to "likely") voters makes them a little better as data tools than 2,587 randomly dialed individuals across the entire country.

I'd love to see Gelman's redo this analysis using exit poll data. When we talk about election results, we should be basing our analysis on election results.

Originally posted to Daily Kos on Thu Mar 26, 2009 at 08:08 AM PDT.

EMAIL TO A FRIEND X
Your Email has been sent.
You must add at least one tag to this diary before publishing it.

Add keywords that describe this diary. Separate multiple keywords with commas.
Tagging tips - Search For Tags - Browse For Tags

?

More Tagging tips:

A tag is a way to search for this diary. If someone is searching for "Barack Obama," is this a diary they'd be trying to find?

Use a person's full name, without any title. Senator Obama may become President Obama, and Michelle Obama might run for office.

If your diary covers an election or elected official, use election tags, which are generally the state abbreviation followed by the office. CA-01 is the first district House seat. CA-Sen covers both senate races. NY-GOV covers the New York governor's race.

Tags do not compound: that is, "education reform" is a completely different tag from "education". A tag like "reform" alone is probably not meaningful.

Consider if one or more of these tags fits your diary: Civil Rights, Community, Congress, Culture, Economy, Education, Elections, Energy, Environment, Health Care, International, Labor, Law, Media, Meta, National Security, Science, Transportation, or White House. If your diary is specific to a state, consider adding the state (California, Texas, etc). Keep in mind, though, that there are many wonderful and important diaries that don't fit in any of these tags. Don't worry if yours doesn't.

You can add a private note to this diary when hotlisting it:
Are you sure you want to remove this diary from your hotlist?
Are you sure you want to remove your recommendation? You can only recommend a diary once, so you will not be able to re-recommend it afterwards.
Rescue this diary, and add a note:
Are you sure you want to remove this diary from Rescue?
Choose where to republish this diary. The diary will be added to the queue for that group. Publish it from the queue to make it appear.

You must be a member of a group to use this feature.

Add a quick update to your diary without changing the diary itself:
Are you sure you want to remove this diary?
(The diary will be removed from the site and returned to your drafts for further editing.)
(The diary will be removed.)
Are you sure you want to save these changes to the published diary?

Comment Preferences

  •  LOl.. dare I say it (6+ / 0-)

    White Power!

    Obama won all whites: 54-45 percent for those making under $50K, and 51-47% for those making over $50K.

    "All of those thoughts were posted on a vicious far-left website called the Daily Kos" -- Bill "falafel" O'Reilly

    by State Department on Thu Mar 26, 2009 at 08:12:20 AM PDT

  •  Oh Kos, you and your radical lefty ideas. (9+ / 0-)

    When we talk about election results, we should be basing our analysis on election results.

    •  I found the mapping a bit disturbing also (0+ / 0-)

      that polling data was used and not 'clearly' extrapolated.  I suspect that Obama's campaign has some ever better data than that.

      I am still scratching my head about just how people are trying to make the election about race - and it really was about competence.  Do we want competence in the white house or do we want a continuation of deterioration education, lack of health care and exhausting our oil supply?

      I believe that competence is colorless.  But let's run the numbers. But I could be wrong, I was surprised when Bush won 2 times.

      "Make no small plans. They will fail to stir humanity's blood, and will not be built anyway." - Daniel Burnham

      by mollyk on Thu Mar 26, 2009 at 08:27:17 AM PDT

      [ Parent ]

  •  So what's the bottom line nationally? (4+ / 0-)

    Did Obama win a majority of all whites, or a majority of those in lower incomes, or what?

    I need to know, because he's not really president until he can close the deal.

    Congress! Pass a 100% tax on Bush's pension, because tax dollars shouldn't reward incompetence that blew up the country.

    by Inland on Thu Mar 26, 2009 at 08:14:56 AM PDT

  •  Does that include the multiracial? (3+ / 0-)

    You know, the half-whites?

  •  I think we need to see.... (9+ / 0-)

    ...all these white guys'  birth certificates. :-)

    OVER HERE: AN AMERICAN EXPAT IN THE SOUTH OF FRANCE, is now available on Amazon US

    by Lupin on Thu Mar 26, 2009 at 08:16:22 AM PDT

  •  Does the point about Appalachia still stand? (2+ / 0-)
    Recommended by:
    theran, State Department

    I seem to remember that at the root of this dispute is a question over whether white voters in Appalachia were distinctly anti-Obama as compared to white voters in the rest of the country, and/or the rest of the South.

    •  Best part of that in my mind (4+ / 0-)
      Recommended by:
      lorzie, brklyngrl, pademocrat, shenderson

      was how UNTRUE it was in Virginia and North Carolina, especially since Obama spent time and effort to win both of those states -- the edge of the "appalachia" line in those figures carefully skirts the state lines of NC and VA.

      I find that ... illuminating.

      Happy little moron, Lucky little man.
      I wish I was a moron, MY GOD, Perhaps I am!
      -Spike Milligan

      by polecat on Thu Mar 26, 2009 at 08:25:03 AM PDT

      [ Parent ]

      •  Yes. I feel like we've been round and round (1+ / 0-)
        Recommended by:
        polecat

        with this question, and are still no closer to having an answer. It would be interesting to try to look at voting patterns parceling out the effects of active campaigning. I'd guess that's where most of the action is. Or, at least I'd like to believe at that work made a difference. :)

        •  Look... (4+ / 0-)
          Recommended by:
          polecat, theran, brklyngrl, OnlyWords

          ...here for my discussion on it from a few days ago using the presidential-vote-by-CD data.

          "Dignified people, without a whimsical streak, almost never offer fresh insights, in economics or anywhere else." Paul Krugman

          by Dana Houle on Thu Mar 26, 2009 at 08:40:32 AM PDT

          [ Parent ]

          •  Thanks, I just checked that out. (0+ / 0-)

            I think it's pretty hard to argue that Obama didn't do distinctly worse in Appalachia than other places - he clearly did. I also think it's clear that he did worse among poorer white voters there than he did in many other parts of the country (New Hampshire, Colorado, etc.).

            What I'm still not clear about is whether he did worse with poorer white voters in Appalachia than with poorer white voters in the rest of the South. As in, if we just looked at white voters in Mississippi or South Carolina, would we still see these declines - which to me begs the question of whether this is something about Appalachia or something about Southern whites (like my family! So, I'm interested!)

            •  Declines in Appalachia Are New (5+ / 0-)
              Recommended by:
              sny, polecat, theran, Grep Agni, brklyngrl

              White voters in the Deep South haven't been voting for Dems for a long time; decades really.  But they've continued to vote Dem in Appalachia and the Ozarks; there may be no states as dominated by the Democratic party as Arkansas and West Virginia.  But this election, while continuing to vote Democratic for other offices, voters in those two states--both of which are overwhelmingly white--swung against Obama and gave him less of the vote than they gave Kerry.

              That only happened in 22% of the counties in America.  I didn't happen in plenty of other states uncontested by Obama, such as high-Dem areas in New England (other than MA, where the 2004 numbers were skewed by it being Kerry's home state), the Plains, California or even much of the Deep South.  Where it happened was the arc of Appalachia down through TN, northern AL and northern MS, in to the Ozarks and then in to Oklahoma.  These places were all settled primarily by people whose ancestors came from the border regions of Great Britain, settled in Appalachia, and then if they left mostly settled in TN, in the Ozarks, the Southern Plains (including Oklahoma) and then moved westward in to Southern California (which was until about the last 15 years a fairly conservative place in voting patterns).  

              "Dignified people, without a whimsical streak, almost never offer fresh insights, in economics or anywhere else." Paul Krugman

              by Dana Houle on Thu Mar 26, 2009 at 09:12:02 AM PDT

              [ Parent ]

              •  NC Appalachia is otherwise (1+ / 0-)
                Recommended by:
                Donkey Hotey

                The western part of NC has always been both overwhelmingly white and overwhelmingly Republican.
                That we now have a Democratic oasis in the counties
                surrounding the two campuses of the state university
                system in that region is what is new, for us.

                The road to hell has not YET been paved with Republicans, but it SHOULD be -- Corrected BumperSticker

                by ge0rge on Thu Mar 26, 2009 at 01:07:52 PM PDT

                [ Parent ]

              •  Thanks. I think what I'd been wondering, (0+ / 0-)

                and dreaminonempty has the data for this in a comment just below in this same thread, is whether still further declines among white voters in the Deep South were being masked by the very large numbers of black voters in those states (and in counties within those states).

                Politically, it probably doesn't matter much since we're not going to win Mississippi for a long time. I was just theoretically curious what had happened to the white vote in the Deep South versus Appalachia.

            •  Hi brklyngrl! (1+ / 0-)
              Recommended by:
              brklyngrl

              How ya doin?  I've been quiet for a while - busy - but I think I have the data at my fingertips that may interest you.  Here's a comparison of how Kerry and Obama did among white voters only:

              First, the Appalachia/Upland South/Oklahoma axis -

                Kerry  Obama  Change
              WV 42%    41%    1 pt (3%)
              KY 35%    36%    +1 pt (+3%)
              TN 34%    34%    no change
              AR 36%    30%    6 pts (17%)
              OK 29%    29%    no change

              Next, the Gulf Coast/Deep South

              LA 24%   14%    10 pts (42%) (Katrina?)
              MS 14%   11%    3 pts (22%)
              AL 19%   10%    9 pts (47%)

              At the state level, it's clear that the changes among whites were far more severe in both absolute and relative terms among whites in the AL, MS and LA.  The state level masks Appalachia everywhere except WV, though, so let's look at counties.

              Are these declines in AL & MS just because of declines in Appalachian Alabama and Mississippi?  We have some counties in northern Alabama and Mississippi that are nearly all white that give us a good idea:

              AL
              Cullman    23%   17%  6 pts (26%)  
              Winston    21%   18%  3 pts (14%)  
              Marion     30%   21%  9 pts (30%)
              Blount     18%   15%  3 pts (17%)

              MS
              Itawamba   29%   21%  8 pts (28%)
              Tishomingo 34%   23%  11 pts (32%)

              Answer: No.  Whites in the rest of the state must have decreased their vote share even more than those in Appalachian AL & MS in order to be consistent with the exit polls.

              What about Eastern Kentucky?  In absolute terms, there's some pretty large changes.  Let's look at the most severe:

              Letcher   46%  32%  16 pts (30%)

              Not the worst relative change, and better than the statewide results for both LA and AL (but worse than KY as a whole).  There's only a couple counties in E. KY with changes this large.

              So, in absolute terms, decrease in vote share for the Democrat among whites only was nothing for most of WV, about 5 points for most of Appalachian KY and southern WV, and about 10 points throughout the rest of Appalachia and the Gulf Coast.  In relative terms, however, the change became far more severe the further south you go, and was worst to the South of Appalachia.

              To answer your question directly: Yes, if you just look at white voters in Mississippi, outside of Appalachia, you still see such declines - even worse, actually.

      •  Not "Untrue" (1+ / 0-)
        Recommended by:
        polecat

        I wouldn't say it was "untrue."  It wasn't, however, something that couldn't be overcome and changed.  

        I wrote on this subject most recently a couple days ago.

        "Dignified people, without a whimsical streak, almost never offer fresh insights, in economics or anywhere else." Paul Krugman

        by Dana Houle on Thu Mar 26, 2009 at 08:39:32 AM PDT

        [ Parent ]

        •  Perhaps I didn't chose my word wisely, (0+ / 0-)

          and I did like your article very much.

          But please note how the NC and VA lines are carefully skirted in the CD data.

          That at least gives me hope for the future, assuming anyone wants to dive into TN, KY, and WV and do something about the neanderthals living there.

          Happy little moron, Lucky little man.
          I wish I was a moron, MY GOD, Perhaps I am!
          -Spike Milligan

          by polecat on Thu Mar 26, 2009 at 08:46:07 AM PDT

          [ Parent ]

    •  Yes n/t (1+ / 0-)
      Recommended by:
      brklyngrl

      "Dream for just a second and then do it!" -- Kolmogorov

      by theran on Thu Mar 26, 2009 at 09:53:45 AM PDT

      [ Parent ]

    •  How is that the root of a dispute? (0+ / 0-)

      The point about white voters in other parts of the country being anti-Obama is very well taken, at least
      in the other (than Appalachia) part of the south.
      I did NOT see how there was ANY "dispute" about
      this at all.  Those figures would simply HAVE HAD
      to be that way purely on the basis of the overall
      figures.  Anything less would've meant a bigger landslide.  The population of Appalachia isn't big
      enough to account for the overall state margins that
      were observed in southern states.

      The road to hell has not YET been paved with Republicans, but it SHOULD be -- Corrected BumperSticker

      by ge0rge on Thu Mar 26, 2009 at 01:06:31 PM PDT

      [ Parent ]

  •  A more interesting question (2+ / 0-)
    Recommended by:
    Donkey Hotey, OnlyWords

    would be education levels rather than race or income.  As I show my intellectual elitism here, I think that a stark difference in voting patterns would leap out of the pages...

    No politician ever lost an election by underestimating the intelligence of the American public. PT Barnum, paraphrased...

    by jarhead5536 on Thu Mar 26, 2009 at 08:19:56 AM PDT

  •  Primary reason to cherry pick data (2+ / 0-)
    Recommended by:
    Superribbie, Donkey Hotey

    is to have the data fit the conclusion you already had.

    Don't be surprised if there is not a redo.

    Also don't be surprised if his "study" is used to justify all kinds of things.

    Happy little moron, Lucky little man.
    I wish I was a moron, MY GOD, Perhaps I am!
    -Spike Milligan

    by polecat on Thu Mar 26, 2009 at 08:23:43 AM PDT

  •  Using Pew polling for anything (1+ / 0-)
    Recommended by:
    polecat

    Is iffy at best.

  •  Again, North Dakota (0+ / 0-)

    Obama abandoned this state for strategic reasons, and when he looked to re-enter the state as far as resources go, he again scrapped the state.

    Obama always had a chance, but rightly realized that states like Missouri, Indiana and North Carolina were also going to be close. The thing is, Montana and North Dakota could have not only helped solidify the Western shift, North Dakota would have also helped split the plains (other than Omaha, Nebraska stayed damn red, despite some believing the Senate race was inherently more plausible than ND-Pres)

    That is all. Individually, I wish you the best, but collectively, my dearest hope is to outlive you - groovetronica

    by Nulwee on Thu Mar 26, 2009 at 08:28:07 AM PDT

  •  Is Jindal an honorary white now? (1+ / 0-)
    Recommended by:
    Aprognosticator

    He's earned it.

    Or should he go back to the Bearded Spock universe where he comes from?

    OVER HERE: AN AMERICAN EXPAT IN THE SOUTH OF FRANCE, is now available on Amazon US

    by Lupin on Thu Mar 26, 2009 at 08:28:49 AM PDT

  •  individual-level exit data should arrive soon (3+ / 0-)

    That will be a lot more useful for this sort of analysis than rehashes of the current tables -- although why Gelman used the Pew data, I don't immediately understand.

  •  He could have merged both datasets and weighted (0+ / 0-)

    the underrepresented observations from the polls. Merging both datasets--i.e. exit polls-elections results and pre-election polls--is an acceptable technique. He could have done that and then run his regression.

    Or he could have merged the exit poll data with the census data to introduce an income variable in his analysis and then run the regression.

    There are several ways to get an accurate reading of how folks voted while controlling for race and income.

    Don't give a damn a/t each & every politician currently alive in the US. Last time i voted for the top part of the ballot was 1972. Never missed SB election

    by Mutual Assured Destruction on Thu Mar 26, 2009 at 08:29:44 AM PDT

  •  It's unclear to me (0+ / 0-)

    what the implications of this is.

    •  The Implication is Don't Say You Know... (0+ / 0-)

      ...who voted how by looking at data collected before the election.  

      "Dignified people, without a whimsical streak, almost never offer fresh insights, in economics or anywhere else." Paul Krugman

      by Dana Houle on Thu Mar 26, 2009 at 08:43:56 AM PDT

      [ Parent ]

  •  I was confused by this, too... (0+ / 0-)

    If there exist data that do basically what you want (exit polls), and you instead use different data and make lot of assumptions about underlying distributions to get what you want (which is what he did), you at least have to explain why you didn't use the other data.

    I don't get why he didn't bother to do that.  I mean, I'm sure there are reasons not to use exit polling, but you at least have to mention those.

    Guide to my comments: When in doubt, assume sarcasm.

    by Gray on Thu Mar 26, 2009 at 08:31:11 AM PDT

  •  Wow Kos calling the bluff on fivethirtyeight (5+ / 0-)

    for statistical sleight of hand! Someone just trumped the (supposed) ace.. :)

    But seriously, what Gelman did seems dumb enough for even me to understand that it was stupid.

  •  I looked at the 538 post and said "Huh?".... (4+ / 0-)
    Recommended by:
    vpd4, lorzie, math4barack, roadbear

    at least as far as my home state of Washington is concerned.  I know I live in Seattle and it's a little more diverse, but generally the state is pretty white and generally fairly well off - what's more, the state was one of the last to feel the effects of the housing bubble collapse.  Nevertheless, there was no question that Obama was going to carry the state (as has every Democrat since 1988). There was a period immediately following the Palin pick where I think the white vote might've been in play, but that quickly evaporated and those of us who were involved in the campaign breathed a sigh of relief (not that we ever let up on our efforts to get out the vote in huge numbers).  Under Gelman's rubric I don't see how Obama could've won in the way that he did - logic alone (aside from statistics) suggests otherwise.

  •  I had a different take (1+ / 0-)
    Recommended by:
    pHunbalanced
    1.  Look at the Rich white vote:  Obama cleaned up in the northeast, Illinois (not a surprise), Wisconsin, and in Oregon.

    If the Republicans can't win rich whites, they are screwed.

    1. Look at the overall "poorest of the poor" maps:  solid blue, save West Virginia, Utah, Idaho and Wyoming.  The Democrats cleaned up here.

    When liberals saw 9-11, we wondered how we could make the country safe. When conservatives saw 9-11, they saw an investment opportunity.

    by onanyes on Thu Mar 26, 2009 at 08:37:52 AM PDT

  •  I don't know (2+ / 0-)
    Recommended by:
    pHunbalanced, Susan Leonard

    but it seems likely that people confronted face to face might have lied more than folks in a phone poll.  That's my experience anyway.

    We're all one heartbeat away from Forever. kasandra.us

    by KS Rose on Thu Mar 26, 2009 at 08:51:17 AM PDT

  •  Where is the map of how blacks voted? (4+ / 0-)
  •  How Whites really voted in 2008 (2+ / 0-)
    Recommended by:
    sny, math4barack

    regardless of income, education, etc, etc...

       % White Vote in States Won by Obama
     Ordered by % White Male Vote Won by Obama

    StateWhite VoteState Result
    </th<th>MaleFemale ObamaMcCain
    Wash D.C.8686937
    Hawaii72687227
    Vermont63716831
    Oregon55585741
    Washington55565841
    Massachusetts 53656236
    Maine52635841
    Wisconsin51575642
    California 50556137
    Minnesota 49565444
    Iowa49535445
    Rhode Island48666335
    New Hampshire48605445
    New York 48576237
    Colorado48525445
    New Jersey 48505742
    Michigan47545741
    Illinois 46556237
    Ohio45475147
    Connecticut 44586138
    Maryland 44506237
    Pennsylvania43515543
    New Mexico 43415742
    Delaware 42626237
    Florida 42425148
    Indiana41485049
    Nevada40495543
    Virginia37405346
    North Carolina323849.949.5

    Remember, these are the states that Obama won! Obviously white males wanted Palin\McCain, as in only 8 states, and Wash DC, did the majority of white males vote for Obama. In seven of the states won by Obama - Florida, North Carolina, Virginia, Ohio, Indiana, New Mexico and Nevada - the majority of both white men and white women voted for Palin\McCain. These states represent 97 Electoral Votes, enough to have flipped the election to the republiklans - 270EV's to 268EV's.

    In the 2008 election, whites constituted 74% of the electorate. If the racial demographics of the 2008 election were the same as in 1992, when whites constituted 87% of the electorate, McCain would be POTUS. This is why I say "America - still center-right, just less white"

    After having a POTUS who couldn't watch TV and eat a pretzel at the same time, it's great to have one who can.

    by OnlyWords on Thu Mar 26, 2009 at 09:01:35 AM PDT

  •  Your analysis makes sense to me, Kos. Great work (0+ / 0-)

    and what you wrote seems correct.

    Wonderful job-- thanks for doing that !!

  •  More data - 2004 (1+ / 0-)
    Recommended by:
    Donkey Hotey

    After the 2004 elections, I looked at exit polls to see in which states Kerry won the white vote.  

    The results, in no particular order: - The New England states (not including NY) - Minnesota - Washington - Hawaii
    Oregon was a 50-50 tie.

    Kerry lost the white vote by 4 and 1 points respectively in the two biggest blue states (CA,NY).

  •  . (2+ / 0-)
    Recommended by:
    lorzie, baudelairien

    Sorry, Kos, but this post is absolutely wrong.
    Exit Polls don't have a sample size of 8,000 per state. I don't know from where you get that idea. It's more like 1,250.
    There- Exit Poll choose a medium-sized state like Alabama, Arizona or Kentucky- the sample size is always something like 1,000 to 1,500. This number is higher for big states and Toss-Up states and lower for small, uncompetitive states (like Idaho and Hawaii- they're more in the 800 range).

    So- what makes you think that a random telephone poll is preferable to a live-interviewer poll on election day? A telephone poll is exactly that- random. If a reputated institute like Pew does the poll, there is a good chance it will be representative.

    What is an exit poll? Some researchers stand in front of just a few precincts in the state and ask all people if they want to participate in a poll.
    If there is no exit pollster in a rural Texas county, there is no way that any voter from that county will be representated in the poll.

    Los Angeles County has almost 5,000 precincts.
    The exit poll for the WHOLE STATE OF CALIFORNIA had 2,200 respondants. There is no chance that this poll can be right, except for random luck. Remember the 2004 NC exit poll that had black voters at 26%? Well, the Secretary of State office reports the real numbers. How many black voters were there? 19.5%. Great poll. I'd prefer a Pew or Gallup poll over an exit poll any day, if I had the chance to use both... Well, the exit polls are free, the Pew data not, so I don't have a choice but to use Exit Poll data, but I'd love to have access to Pew.

    So... maybe you should think about your criticism of Gelmans analysis again, because what he did is a lot more solid than this post.

    •  Wow (2+ / 0-)
      Recommended by:
      math4barack, c liam

      Los Angeles County has almost 5,000 precincts.
      The exit poll for the WHOLE STATE OF CALIFORNIA had 2,200 respondants. There is no chance that this poll can be right, except for random luck.

      Well, using your reasoning, no poll could ever be accurate.  

      "Dignified people, without a whimsical streak, almost never offer fresh insights, in economics or anywhere else." Paul Krugman

      by Dana Houle on Thu Mar 26, 2009 at 09:15:04 AM PDT

      [ Parent ]

    •  Listen, exit polls are polls of people after they (0+ / 0-)

      already voted. So, you don't have to wonder if they really will vote...

      And you do not appear to understand the formula used to generate margin of error either.
      1/ square root of n.  

      And you can weight around sampling bias.  

      There's also bias in who will answer the telephone as so many of us have caller id and others don't have landlines.

      I agree with Kos.

      •  There is actually a difference between (0+ / 0-)

        the margin of error of a random sample, as in a poll, and a non-random sample, as an exit poll.
        Dana, of course a poll has a chance to be perfect, because most people can be included in the poll. Pew even uses cellphones, as far as I know.
        But in the Exit Poll, something like 95% of voters had NO CHANCE to participate because there was no exit pollster near their precinct. 2,200 respondants, probably from less than 500 precincts, while there are about 20,000 precincts in California.

        The exit poll is a (more-or-less, even here apply caveats, like- did the pollsters do interviews the whole voting time?) random sample of those 500 precincts, and the margin of errors applies for those. And I think that Edison&Mitovsky tries to choose the precincts as such that they are as representative of the State of California as possible (some rural areas, more urban areas, some GOP leaning areas, some Dem-leaning areas etc.pp)... but that is impossible when you don't do a random telephone poll.

        math4barack- I know the formula for MoE, and I know what an exit poll is. But the Margin of Error is used for expressing the confidence interval of RANDOM sampling errors. The sampling errors of the exit poll are not random, however, they get that errors because their sampling poll includes just like 5% of the voters. Not their sample- the pool they draw the sample from.

        Of course, telephone polls have their own problems- caller ID, likely voter model errors (however not cellphones, because Pew polls those)- but they diminish when you look at the problems of the exit poll.

        •  . (0+ / 0-)

          however, they get that errors because their sampling poll includes just like 5% of the voters.

          Sorry, I mean the sampling pool, of course.

        •  the precincts arn't random though (0+ / 0-)

          The precincts for exit polling are very carfully selected to reflect the state as a whole.  Precincts that have similar results to the state as a whole, precints that can predict turnout for both republicans and democrats.  It's a bit of a science, and is why they can use a small sample to estimate results.  I once say someone estimate, within a couple percentage points, the results from a city election with only only one precinct results.

          Also, a  sample size of 2500 isn't small for the state of CA.  Many national polls use a sample size of 3000.

  •  The Irony is Deafening (3+ / 0-)

    I'd love to see Gelman's redo this analysis using exit poll data. When we talk about election results, we should be basing our analysis on election results.

    Exit polls are no closer to election results than Pews polls taken prior to the elections. Neither are known to be representative of actual votes.

    Granted, you can make good arguments why one method of polling is better than another. But there's no need to pretend like one form of polling is something other than what it is, polling.

  •  I have to ask myself (0+ / 0-)

    Who pays this guy to generate (mis)information that is only useful to Republicans?

    Or did I just answer my own question?

    •  Columbia University Does (0+ / 0-)

      "Dignified people, without a whimsical streak, almost never offer fresh insights, in economics or anywhere else." Paul Krugman

      by Dana Houle on Thu Mar 26, 2009 at 09:33:21 AM PDT

      [ Parent ]

    •  he wanted to use individual-level data (0+ / 0-)

      and this is what he could find. I wouldn't have gone there, but I don't see how it has anything to do with partisan axe-grinding.

      •  I Don't See Partisanship at Play (3+ / 0-)
        Recommended by:
        theran, cskendrick, HudsonValleyMark

        But I see shoddy work.  His previous post was a joke.  He asserted that there was nothing going on regarding Appalachia, contrary to what I've written.  But he simply showed two maps, that weren't tracking the same thing.  He got drilled pretty hard in the comments, but never really addressed the (imo valid) criticisms raised.  

        I agree, not partisan axe-grinding.  But it makes me doubt his credibility on other subjects, seeing how shoddy these two post have been.

        "Dignified people, without a whimsical streak, almost never offer fresh insights, in economics or anywhere else." Paul Krugman

        by Dana Houle on Thu Mar 26, 2009 at 09:45:25 AM PDT

        [ Parent ]

      •  I Assume You Probably Saw... (2+ / 0-)
        Recommended by:
        theran, HudsonValleyMark

        ...that early post I mentioned, but if not, it's here.  

        And the problem, again, is that he showed two maps, one with overall vote changes by county, then another with his extrapolations of the white vote by county in 2008, and says "nothing to see here," but didn't look at any possible changes in the white vote between 2004 and 2008.  

        Classic case of apples and oranges.  

        "Dignified people, without a whimsical streak, almost never offer fresh insights, in economics or anywhere else." Paul Krugman

        by Dana Houle on Thu Mar 26, 2009 at 09:49:46 AM PDT

        [ Parent ]

      •  He had access to the raw data? (0+ / 0-)

        Alright, that would justify the recategorization of income bands, absent data sparseness issues.

        But testing cells by income bands and state of residence?

        I think that calls for a population study or very large sampling, on the order of tens of thousands of respondents.

    •  That's offensive (3+ / 0-)
      Recommended by:
      theran, baudelairien, arainsb123

      Look, I acknowledge that there maybe flaws in this particular study.  But there's no need for the personal attacks.  Nate, one of last year's biggest progressive superstars, has invited Gelman to contribute to his site because he's one of the most accomplished social statisticians in the country.  His book from last year was glowingly reviewed across the left blogosphere, and serves as one of the best political-geographic snapshots of the electorate.  Leaping from that to the implication that he's doing hack Republican science seems wholly unnecessary.  Let's get Peter (plf515) or another dKos statistician to look at it, and then maybe we'll get somewhere.  

      Your Texas bashing makes you sound stupid.

      by cardinal on Thu Mar 26, 2009 at 09:41:05 AM PDT

      [ Parent ]

      •  Goes to market for this type of information (0+ / 0-)

        Choices were made to

        1. study the 2008 election for white participation
        1. using data of a type and structure that heavily informed the results
        1. in a fashion that amplified an uncontroversial conclusion to a controversial extent
        1. Without testing against other sources of data nad structures

        This guy might be the finest statistician in the world.

        And he still fell flat on this one exercise.

  •  A more accurate way to analyze voting results (1+ / 0-)
    Recommended by:
    theran

    is to look at actual election results by county and apply the demographic data for the county's voting population to give you a baseline and then make adjustments by exit poll data (which can account for turnout, voter group intensity etc.).

    I think one will find that Obama won the surburban white vote in all of the states that he won and lost the rural white vote at about the same rate as the typical Democrat not named Bill Clinton in most (but not all) states.  

    In the states he won by double digits, he probably won a majority of the white vote.  In states that he won narrowly, he probably lost the white vote, but won decisively among non-whites to earn a victory.

    Alternative rock with something to say: http://www.myspace.com/globalshakedown

    by khyber900 on Thu Mar 26, 2009 at 09:40:23 AM PDT

  •  it would be great (0+ / 0-)
    to use actual results, but unfortunately the only comprehensive "results" we have are the vote... which isn't broken down by race or income category.

    exit polling is notoriously unreliable, especially for subgroups.

    didn't we already have this conversation about a million times over proposition 8?

  •  So are you saying the GOP should pay white people (0+ / 0-)

    more money?

    LOL

  •  Maybe NH has no poor white folk? (0+ / 0-)

    Maybe they exported them all like they rounded up the homeless in a Minnesota before the GOP had their big party in town.

    Just kidding. But it is an odd demographic twist for NH.

    "I think it's the duty of the comedian to find out where the line is drawn and cross it deliberately." -- George Carlin, Satirical Comic,(1937-2008)

    by Wynter on Thu Mar 26, 2009 at 10:50:21 AM PDT

  •  Connecticut Lowest Quintile Not a Mystery (1+ / 0-)
    Recommended by:
    Donkey Hotey

    "...He has Connecticut red for the under $20K white vote, while Obama won the under-$50K white crowd by a solid 59-36. Looking at the income and race exit poll data in that state, it seems improbable that Obama would've lost the under $20K crowd..."

    Actually, I think that a good case exists for Obama losing the <$20 group in CT.  In CT, I suspect that most white male voters with that level of income are retired.  Further, CT also has a number of mid-old and old-old (which makes them members of the "Greatest Generation" and WW2 veterans) voters still living in it.  That group would have had a very difficult time voting for Obama.  Throw in the situation that younger white males making <$20K also tend to have unsettled living situation (read: not registered to vote) and the electorate becomes much more interesting.  Since retirees make up a pretty decent slice of the CT electorate, a very good case could be developed that the <$20K electorate selected McCain, a fellow veteran.</p>

    "Love the Truth, defend the Truth, speak the Truth, and hear the Truth" - Jan Hus, d.1415 CE

    by PrahaPartizan on Thu Mar 26, 2009 at 11:00:23 AM PDT

  •  OK, so Gelman's an idiot (0+ / 0-)

    Telephone polling in advance was discredited since
    Dewey/Truman in 1948.  All I want to know is, is this smackdown going to MAKE IT BACK to any distribution
    point where Gelman will suffer the humiliation HE ACTUALLY DESERVES FOR HAVING DARED this idiocy?
    That's what I WISH I knew how to cause.

    The road to hell has not YET been paved with Republicans, but it SHOULD be -- Corrected BumperSticker

    by ge0rge on Thu Mar 26, 2009 at 12:59:26 PM PDT

  •  This post seems more flawed than Gelman's (0+ / 0-)

    While elements of Gelman's analysis are clearly flawed, this diary seems to take the populist though ignorant view that exit polls are better than telephone polls because they poll people after they actually voted. This ignores all the huge problems that we know exit polls have in terms of things like non-response bias. In fact, there is no reason whatsoever to think that an exit poll is more accurate than a good telephone poll done shortly before the election. Does anyone remember 2004, where the exit polls gave Kerry a 3-point win nationally?

  •  So Much for the Alleged "Appalachian Problem" (0+ / 0-)

    The diary results for Ohio prove what I've been saying all along -- that support for Obama among Ohio whites was directly related to income, with the POOREST whites voting most strongly for Obama.

    Those poor whites in Ohio are largely located in the 32 Appalachian counties. In other words, if the supposed anti-Obama sentiment in Appalachia were correct, as repeatedly insisted by DHinMI, then Ohio would have been among those states where low-income whites voted for McCain.

    That so-called Appalachia problem is a bunch'o'bull.

    •  except that... (1+ / 0-)
      Recommended by:
      TrueBlueDem

      methinks Tuscawaras is the only arguably Appalachian county in Ohio that Kerry lost and Obama picked up. Most of those counties did vote for McCain. You could look it up.

      Of course income was a strong predictor of vote choice. That isn't enough to settle the Appalachian issue one way or another.

      •  wrong metric (0+ / 0-)

        In more than half the Appalachian counties, Obama did better than Kerry, even if still losing. For example, Obama won the city of Chillicothe, which Kerry lost, and Obama lost my county, Pike, by only 170 votes. It was because Obama did so well in Appalachian Ohio compared to Kerry that Obama won the state.

        The argument that Appalachia was somehow genetically anti-Obama is simply wrong.

        •  whoa there (0+ / 0-)

          It was because Obama did so well in Appalachian Ohio compared to Kerry that Obama won the state.

          Because he did better in more than half those counties? You think that created the 6.6-point swing? This I doubt.

          Kerry lost Pike by only 530 votes, so that isn't much of an example: Obama did reduce the margin by 3 points, but that still underperformed his statewide performance. Next county over, in Jackson, basically a dead heat (i.e., deeper underperformance).

          I see 11 counties in the state where Obama did worse than Kerry, and 9 of them are on the east/southeast border. I'm happy to hear your explanation.

          The argument that Appalachia was somehow genetically anti-Obama is simply wrong.

          I'll grant you that, but I don't think that's the issue.

          To get some sense of what the rest of us actually are talking about, try heading over to Leip's site, clicking on Ohio, then selecting the "swing" map. (Or for a more dramatic contrast, try the "trend" map.) Most of the big swings to Obama were as far from Appalachia as one could get.

          Obviously these data won't support some cartoon argument about how denizens of Appalachia would sooner die than vote for a black man. But I don't see how they can be made to support your blanket dismissal of DHinMI's argument, either.

          (As for whether low-income whites in the Appalachian counties literally favored McCain, I can't tell. But the argument that low-income whites favored Obama, and low-income whites are "largely located" in the Appalachian counties, ergo... well, I'm not quite sure ergo what. It's a bit of a mess. You seem to be making the case that Obama must have done best in Appalachia -- although, patently, he didn't.)

          •  It's not as unclear as you think (0+ / 0-)

            You say that Obama underperformed Kerry in 9 Appalachian counties. That means that Obama outperformed Kerry in 23 Appalachian counties, which I believe does definitively refute DHinMI's contention that Obama had some "Appalachian problem."

            If Obama outperformed Kerry in any other area, for example in Vermont, or in Florida, we would not designate those areas as "problem" areas.

            The nine counties where Obama underperformed in Appalachia are not at all random. They have a striking common characteristic -- they are all in Ted Strickland's old congressional district, or the district he had before the 2000 redistricting. Since everyone in Ohio validly expected that Strickland would have been Hillary Clinton's running mate, the loyalty to Clinton on Strickland turf was intense, and it spilled over to punish Obama in the general. That was most apparent in Scioto County, where Obama underperformed Kerry by about 2,000 votes, an amazingly high number. Scioto is Strickland's home county.

            Everyone knows that McCain had only one chance to win Ohio and that was to achieve a blowout in Appalachia, as Clinton achieved in the primary. Not only did McCain not achieve it, but he fared worse in the region than Bush had done. And that is why Obama won the state -- by not getting blown out in the key swing region.

  •  Why not use precinct data? (0+ / 0-)

    it's publically availible, data exists to estimate the racial and economic makeup of them, and you have exact results, more accurate then exit polling data, which only polls people willing to be polled.

  •  kos, do what you're good at (0+ / 0-)

    and leave the statistics to the statisticians.

  •  Political Beliefs and Behaviors (0+ / 0-)

    I agree with you when you say that early telephone polling/random-digit dialing is bad way to base a study of how white people voted because it isn’t very accurate. Random-digit dialing is conducted by placing calls to phone numbers with randomly chosen exchanges around the country. This method is a bad way to measure public opinion because about seven percent of the population currently doesn’t have a phone. Also many people don’t like to receive phone calls from strangers and hang up the phone on the people conducting the polls. Another reason why Gelman’s data isn’t accurate is because the data that he based his study on was taken before the elections occurred. Peoples’ opinions could have change from the time he took the data, to the time that the election actually occurred. Lastly, how can Gelman know whether the people that he questioned actually voted on the day of election or not?
    Exit polling is when voting places are randomly selected around the nation and workers are sent to ask every tenth person how they voted. This method of surveying people is more accurate because it reaches the people who actually voted and it is also taken as elections are occurring.

Subscribe or Donate to support Daily Kos.

Click here for the mobile view of the site