Wednesday, September 05, 2007


From Rasmussen Reports' General Election Match-Ups, here's what the polls were saying as of Labor Day. (hat tip jamess)

DemocratRepublicanWins by
Edwards 49%Giuliani 41%Dems Win +8%
Obama 44%Giuliani 43%Dems Win +1%
Clinton 44%Giuliani 47%Dems Lose -3%
Edwards 50%Huckabee 33%Dems Win +17%
Obama 48%Huckabee 39%Dems Win +9%
Clinton 49%Huckabee 41%Dems Win +8%
Edwards 45%McCain 38%Dems Win +7%
Obama 46%McCain 40%Dems Win +6%
Clinton 46%McCain 44%Dems Win +2%
Edwards 52%Romney 36%Dems Win +16%
Obama 47%Romney 38%Dems Win +9%
Clinton 51%Romney 40%Dems Win +11%
Edwards 49%Thompson 35%Dems Win +14%
Obama 46%Thompson 39%Dems Win +7%
Clinton 48%Thompson 44%Dems Win +4%

Here's a graphical representation of what that all means. (Click image to see full-size version.)

Now, while some will argue that all the national polls are good for is name recognition, there are some things we can tell from this. While you could make the argument that Edwards has some residual higher name recognition from 2004, with the media's fascination with Obama, they have pretty much equal name recognition. In fact, Gallup confirms this.

With Edwards and Obama sharing almost identical name recognition nationally, and just a bit behind Hillary, there's something going on here that's more than just name recognition when it comes to Edwards' leads in the polls.

Now, the STATE polls are a much different matter. See, Iowa is already (and has been for a few months now) been inundated with radio and TV and newspaper ads from the various campaigns. This is not happening in, say, California. If a voter in Iowa doesn't know who the main candidates are by now, with the caucus possibly being held in December 2007 the way things are going, I'm gonna go out on a limb and say that that person isn't going to show up to vote in the caucus anyway. And FYI, Edwards hasn't run any campaign commercials in Iowa this cycle yet. The only one to air in Iowa was the one with Iowans asking Congress to stop capitulating to Bush on Iraq.

And as for Iowa, here's the latest TIME poll, which was taken before Edwards picked up those key labor endorsements from the Steelworkers and the Mineworkers.

A new TIME poll of likely Iowa caucus goers, taken a week after Edwards' seven-day, 31-stop bus tour of the state, gives Edwards 29% of the vote, five points ahead of Hillary Clinton and seven ahead of Barack Obama. This latest TIME Poll of 519 likely Iowa Democratic caucus goers finds that, among the so-called "second-tier" candidates, Bill Richardson has 11%, with Joe Biden at 5%, Dennis Kucinich with 2%, and Chris Dodd at 1%. (The Poll was conducted August 22 - 25, 2007. The margin of error is approximately +/- 5 percentage points.)

With the field limited to the top four candidates, Edwards's lead over Clinton widens, to 32% to 24%. Obama was at 22%, with Bill Richardson at 13%. Iowa polls can be unreliable, since only 5% to 10% of voters go to the caucuses; some other recent surveys have Edwards in a dead heat with Clinton and Obama. The race remains wide open, but Edwards's position remains strong.

Remember, with the way the Iowa caucuses work, they have a "realignment" once the also-rans are eliminated. So it's very likely the top four will go according to the TIME poll, or perhaps even without Richardson, if he doesn't have enough initial support to qualify going to the next round. Given that Edwards increases his lead over the field when the single-digit candidates are eliminated, Edwards looks to be in good shape to win the Iowa caucuses. And winning Iowa will have an effect on all the other states.

BTW, pollster Mark Blumenthal shares his insight into the differing methodologies of the TIME vs. the American Research Group polls that recently came out. Basically, you have to beware of voter screens that are too wide. Only 5.5% of eligible voters in Iowa showed up for the Democratic caucuses back in January 2004. Those are the hardcore of the hardcore. The real unknown is how much pull Hillary will have with Iowa women who have never participated in a caucus before now. It's one thing to say you're going to show up; it's another thing to actually show up. One truth in polling is that people tend to exaggerate their civic participation; I remember one poll that asked if the person had voted in 2004, and over 75% said they did, when only about 55% of the voting-age population actually did. So if a poll is casting too wide a net, it will most likely be oversampling those with higher name recognition. And I'll add this too. Polls that treat Iowa like a regular primary where you fill out a ballot and go home ignore, at their own peril, the complexities of the caucus system. More polls should ask questions like TIME did, first asking of all the candidates, and then asking a second question of just the top-tier ones, mirroring how the actual process will work.

1 comment:

gabe said...

haha, its almost as if you're a statistics phd student... i know we disagree on this, but i cannot state how meaningless i think polls are at this point- especially ones like this. clinton v huckabee?? who's heard of huckabee? he hasn't even been campaigning outside of iowa. campaigns change everything, and polls at this point are merely name id and party affiliation. polls at this point in '92 were a virtual guarantee that we'd have 4 more years of 41, and look how that turned out.