Above : Kermit goes on Colbert to talk about the primaries. But in 18 days, we’ll see Kermits begin hatching all over the place.
The Gallup Poll is raising many eyebrows. Jay Cost says “I am done with Gallup” after they changed their methodology eight weeks before the election. But Cost also says, “I was a poll junkie in 2008. In 2012, I’m clean.” So he’s not paying as much attention to polls this time around.
Everyone is forgetting this but back in April, Gallup had Romney ahead.
In April, Axelrod tweeted that a poll showing Mitt Romney with a 48-43 percent lead over Obama was “saddled with some methodological problems,” directing his Twitter followers to read a National Journal story criticizing Gallup polls showing a Romney lead.
In that National Journal piece, Ron Brownstein wrote that the polls showing Romney leading the president had “a sample that looks much more like the electorate in 2010 than the voting population that is likely to turn out in 2012.”
What was 2008 more like? 2006 or 2004? It was 2006. So what reason wouldn’t the electorate of 2012 be like that of 2010?
More details of what Gallup was posting, Axelrod and the Justice Department suing the polling firm, and how Gallup responded can be found here.
I’ve been reading around the web and seeing people say that Gallup is different because of its LV model (Likely Voter) in that it polls whites at around 79% (in 2008, I think the percentage of the white electorate was 75%). But remember, LV model doesn’t mean the demographics of the nation. It means who is likely to show up to vote on election day. Gallup is saying that the non-white electorate will not come out in 2012 as they did in 2008 or, equally, that more whites will come out in 2012.
“Is the Gallup polling good? Or is it bad?” Hell if I know. I agree with Cost and just rid yourself of the poll watching. A better idea is to look at the trends in the absentees and early voting. The thing is that Gallup is the august poll and is considered the standard. Healthy cynicism is always good.
The only stupid thing people are doing are looking at PPP or Marist polls. They’re considered so off as to not be biased but to be classified as science fiction and fantasy. Here is what Jay Cost has to say about them in his Twitter:
Marist is a generally lousy and unreliable pollster that’s been promoted wrongly to the big leagues by NBC/WSJ. Any questions?
Increasingly, my response to the proliferation of crap polls is to take the median, rather than the arithmetic mean.
Also, everybody should note well: NBC/WSJ/Marist will likely point to an Obama Electoral College victory UNTIL THE VERY END.
Based on where Marist was the last two times the GOP won elections — 2004 and 2010 — I have no reason to expect otherwise if Rom wins.
Mark my words, Twitterers: on 11-6 the Marist polls will point to an Obama victory.
There are some really fine pollsters out there who have been effectively benched. And we get PPP and Marist day in, day out?!!
THAT explains why President Obama has spent so much time in IA. He’s only leading there by 8 points! Gotcha. Thanks, NBC/WSJ/Marist!!
National polls are very important because they are polled more frequently than state polls. Detections in trends would first be indicated by the national polls.
“But it isn’t a national election. It is fifty state elections.”
While that’s true, national trends are a factor. Consider the states and their electoral votes to be the uneven elevation at the bottom of a pond. Only one percentage point switching for a candidate nationally is enough to move several states over. Let’s say the seven point divide is true with yesterday’s Gallup (52-45) [I don't think that is what it will look like on Election Day, but let us say it is true for this point.] The election result would look like 1988 of Bush crushing Dukakis. Bush had 53% and Dukakis had 46%
I do have an observation to point out about younger people I see on the message forums. They keep talking about close elections, even electoral vote counts, and even someone winning the popular vote and losing the electoral vote (which has only occurred four times in the history of the US).
My observation is that they tend to have a difficult time imagining a big decisive Republican win. “That is because they are leftists.” No, I don’t think that is it. When was the last time Republicans won the presidency decisively? 2004 and 2000 were close. 1996 and 1992 were Democrat wins (though Clinton couldn’t get half the vote due to the strong Perot third party challenge). You have to go as far back as 1988 where Bush defeated Dukakis. That was 24 years ago. Many of these young people weren’t even born then. You have to be at least 40 in order to remember 1988.
Most presidential elections are not close. Check the history. Out of a hundred years of data, there has been no close presidential elections except for 1960 and of course 2000 and 2004 (and what reason that is, anyone can speculate. It could be something to do with Bush, something to do with a lackluster challenger, maybe that the economy was healthy in both). The odds of 2012 aren’t good compared to the historical pattern. Usually, the electorate swings one way more than the other. If Romney wins, it would likely be a decisive win and not a close win.
The political scientists who have the most credibility in their predictions would be the University of Colorado political professors of Bickers and Berry. They have correctly predicted the presidential winner of every election since 1980 even getting 2000 right of Gore winning the popular vote but losing the electoral vote. That’s very impressive. Apparently, their model does state by state economics but also ties in national indicators as well. Things like where a convention is held is factored (interestingly, they both said that the locations of the conventions made zero difference this cycle). They recently updated the latest economic data in their model which flipped New Mexico. Here is a visualization of what they predict:
Some of the states don’t make much sense. Reagan never won Minnesota (but won every other state) so I doubt Romney would carry it. But I’m surprised that New Mexico would go Romney but Nevada would stay Obama. Wouldn’t Nevada flip before New Mexico? Oh well.
They end their study by saying, “While the model has not been incorrect yet, is this the year that it will be wrong?” They don’t know. But they made their prediction and put it up for everyone to see.
I am curious how people can put stock in Nate Silver (who is being outed as being a hack) but dismiss a model that’s been consistently correct. If the prediction proves correct, as it has done for decades, should we not elevate those U of C professors as geniuses? They apparently have figured something out that the rest of us missed.