His own electoral map projection doesn't support his findings of a landslide, but his experience from making lots of similar electoral map projections in the past does. Huh. If he really sees the trending of the data heading toward McCain in so many states he lists as "blue," why list them as blue? Why not make a bunch more of them purple? *shrug again* Meh, I guess we'll see if he's right in a little over a month.
Can you post a link for this? I don't doubt you are telling the truth. I'm just kind of curious to learn more about this.
What I've heard from a couple of pollsters--one being interviewed on NPR the other on Bloomberg--is that Obama needs to be at 50% or at worst 48% in the swing states he wants to win. If he's below 48% he'll lose even if McCain is around 44%-46% because the independents are going to swing to McCain overwhelmingly. I was surprised to hear the same analysis from two different pollsters on two different shows.
I've heard that too, but I do wonder if there's something of an echo chamber effect. One pollster hears another pollster say that, it sounds safe, so he says it too and pretty soon it's a unanimous opinion? I mean, is there really any way to look into the heart of somebody you're polling to calculate how likely racism will play into their vote for US President when you are left all alone at the ballot? Especially when sexism and ageism (and Bidenism!) also all probably have some play too? I dunno. I've heard it so many times that I've come to accept the "3% black buffer rule" (just made that up). But I also really bought in to the "all these polls are under-reporting young cell phone users" line that everyone echoed last time. Hell, maybe he needs a 6% buffer. In which case he's probably fucked.
http://www.fivethirtyeight.com/search/label/pollster ratings The list I posted from surveyusa is from February though, because I followed a link off 538. Still, Zogby has the same rank in the updated one regardless. http://www.surveyusa.com/index.php/...-2008-contests-mean-avg-error-through-050608/
gracias! dammit. I promised myself I wouldn't pay attention to polls until the last week, and I've been sucked in.
Pat Cadell, a former Democratic Party pollster, was on Fox tonight, and his take is the same as Zogby's. He talked about the structure of the electorate and campaigns. He also said Obama is absolutely not handling the banking finance issue well at all. A big part of the structure he talked about is that McCain is seen as a "doer not a talker" while Obama is seen as a "talker not a doer." Pretty fair way to put it, if you put partisanship aside. In his words, Obama basically told congress and the president that he could easily be reached by phone if he was needed - not a good message. It is 20+ years later, but I'll go back to my Harold Washington story. In Chicago, he defeated the current Mayor Daley and current Mayor Jane Byrne for the Democratic nomination in the primaries (in a close 3-way race). In a city that elects democrats as mayors with 70% of the vote, Washington got just 52% of the vote against a nobody/cannon-fodder candidate named Bernard Epton. To my knowledge, Epton was a lawyer in some downtown law firm with no experience and hasn't run since. Mook - if you take into consideration that 18% kind of swing in a modern Northern City (Obama's home town even), and 25 years of progress (hey, we do have a black man nominated with a real chance to win!), maybe your 6% or 3% figures are about right. It sucks; I don't think his race should be a consideration at all. What should be a consideration is that he's basically the same candidate as John Kerry with better ability to read from the teleprompter and a wonderful speaking voice. Kerry's policies hardly wowed anyone.
I might have to come out with my own electoral map this weekend on the PE board. I did pretty well in the primaries.
Here is how Nate Silver said the undecided votes are likely to break: He also had a recent post on the Bradley Effect http://www.fivethirtyeight.com/2008/09/bad-math-and-bradley-effect.html
I think he's using poll data that's a combination of what he sells and we don't see, and questions other than "who would you vote for?"
The outliers only magnify the idiocy of averaging a variety of polls together as if that result means anything. Why is it useful if you average 2 polls where one is a sample of 500 people and a margin of error of +/- 5% and the other is a sample of 1500 people and a margin of error of error of 2%. Or one is a sample of registered voters and the other isn't. Or one is weighted and the other isn't. It's a weak and pathetic attempt to get some sort of consensus among the various polls and techniques, but it has zero mathematical value to it.
I forgot to add this: In fact, the electoral college projections are equally bogus if you have different polling organizations and techniques used to give Obama a certain state vs. McCain another. This is why Zogby's assessment carries so much weight - he's using his own data, consistent techniques across all states he polls in, factoring in sample sizes, and knows from experience that it's more accurate to use results for likely voters (using his techniques).
It'll be interesting to see how Obama's ground game helps him as well, something the polls don't always pick up, and probably aren't picking up. Here in Missouri, you have the McCain ground game as 2 people sitting at a table with NOTHING. Just there to talk. The Obama ground game is 2 people sitting at a table with campaign stuff (stickers, posters, etc.) AND tons of voters registrations forms...not only that, but they have an army of about 20 people at a time going around with clipboards with voter registration forms. In Indiana, they are doing that at gas stations. Obama could easily get a ground game bounce that the polls aren't showing.
I'm calling the Zogby Electoral Map bullshit. I was going through the pollster data, to find the last poll that McCain led in. If you go back, it has been 32 legit polls since McCain led in PA, in April. I say legit, because if you go back 12 polls, you have a McCain lead...done by Zogby. This is one of Zogby's internet polls, which have proved to be completely unreliable (see the 538 pollster ratings). Match the numbers...49 to 44 on pollster.....49 to 44 on Zogby's electoral map. This nutjob is using his internet numbers on his electoral college map. Numbers match up on Colorado saying it's one of his internet polls too. Same thing happens in Florida, and Virginia, and New Hampshire, etc. etc. This Zogby guy's electoral map is complete garbage since he is basing it on polls with internet only samples.
Zogby has a pretty poor reputation among pollsters, mainly because he went way out on a limb to call it for Kerry in '04 and ended up looking like an ass. He also continues to try to claim that internet polls are legitimate, when they have terrible results. (The internet polls are the ones called "Zogby Interactive" that are ranked dead last on BG7 Lavigne's list above.) Whether or not we think Zogby is an especially credible source, I really have a hard time seeing how McCain can win in an electoral blowout. The state-by-state map just isn't favorable for it, and the numbers just haven't been very volatile throughout the campaign. Obama looks to have about 250 electoral votes more or less locked in, McCain about 175-200. That means that McCain pretty much has to run the table of battleground states to win. I think it's quite conceivable that he could do that and end up winning this thing... but with 270-280 electoral votes. If all the tossup states go Obama, then we'd see an electoral wipeout, with Obama getting 350+. Long story short, I think both candidates still have a good chance to win, but only Obama really has a chance to win in a blowout. Of course, I'm not a pollster, even a crappy one like John Zogby I think Nate Silver's projections over at fivethirtyeight.com are by far the most compelling of this cycle. Will be interesting to see if his results compare well to the actual election. It would be pretty cool for a guy who started modeling baseball stats to end up being one of our best political forecasters. SR
You misunderstood my use of "average." I don't mean a simple average, of adding up all the poll margins and dividing by the number of polls. I meant a weighted average. A poll of 1 person would have such a tiny weighting, it wouldn't have any effect on a weighted mean. A poll with a 100% margin of error would have such a tiny weighting, it wouldn't have any effect on a weighted mean. So, outliers absolutely should be counted in. If it's an outlier due to being extremely uncertain (high margin of error) or not polling many people, then it's low weighting will reduce its effect on the "average" in proportion to how uncertain its sample is.
One last thought on the Bradley Effect--ask yourself a question (I got this from Scrubs, but it sort of applies): If you had a choice for a surgeon, and they were equally qualified in every way, would you go with the white woman or the black man? It's a pretty inflammatory question, and I don't want to sidetrack the discussion. But I immediately answered "the black guy," and so did my (also white) wife when I just asked her. I asked her why, and she gave a shy grin and said, "It's probably a sexism thing." I couldn't help but agree. At least in my two person experiment, sexism outweighed racism. I think it's one of a few reasons that torpedoed Hillary's once-inevitable nomination. A lot of people like Sarah Palin (although less than a few weeks ago). They have positive feelings about her. But when they step into the booth and consider that she's a 72-year-old ticker away from being the President, is that going to outweigh racist misgivings? I have no idea. I'd like to see Zogby, or any pollster, explain the formula for calculating that one.
I've often heard people say "A black man will be elected president before a white woman." Intuitively, I'd have guessed racism outweighs sexism, but many people feel sexism is actually much more ingrained and pernicious.
It seems like if you (Staff writer Jill Terreri) are going to say that Zogby thinks the race "will end in an electoral landslide" you should quote Zogby saying that. I suspect that Zogby said something closer to the headline, that it "could" end in a landslide. barfo
