There's been talk about a recent poll which claims to have found it was a mistake to have to gone into Iraq.

Ever being the curious sort, I decided to have a look at the poll itself -- undoubtedly read and believed by unquestioning millions -- before it passed into the collective unconscious. And you'll undoubtedly be shocked -- shocked, dear reader -- at what I've found.

The poll is described by Ipsos here, the actual questions and answers can be downloaded here.

First, let's note that the poll of 1000 people included 50% Democrats and 40% Republicans. In other words, it polled 25% more Democrats than Republicans, so we can see they're clearly oversampling liberal opinion. (Amazingly, the poll claims a 3% error rate. Unless the Democratic Party has seen a headlong rush for membership since November, there's a lot more error here than 3%!)

So when we ask if the Iraq was was a mistake, we get:

Made the right decision in going to war in Iraq	42%
Made a mistake in going to war in Iraq	53%
Not sure	5%

Well, isn't that interesting! Remember we had only 40% Republicans -- and yet 42% of respondents thought we did the right thing. Remember also that 50% were Democrats -- and again, note that 53% thought we made a mistake. Although we can't see the specifics, if we assume equal numbers of cross-overs, it looks as though this poll were being decided by 50 "independents" -- as though you could assess the mood all independents by asking only fifty people!

Also note that if you compensate for the skew in the poll itself, this result comes out to be about dead even between the two answers. This also throws Bush's "approval" ratings in the poll into context, where the poll claimed 41% approve, and 56% disapprove, revealing a slight (though only slight) edge for disapproval -- not the 14% difference this poll claimed to have found.

But then again, you can find almost anything when sample 25% more Democrats than Republicans. Hmmmm... I wonder why the "error" always seems to go that same way?

The poll of 1,000 adults was conducted June 20-22 for the AP by Ipsos...

So the poll was comissioned and paid for by the AP itself. Imagine that. Interestingly, this poll apparently generated two press releases, one saying that people oppose the draft (no kidding -- so who was in favor of it?) and then a second one in which they announced that it showed less support for the war than before.

Strangely, there was no press release for a third finding -- that most people didn't want our troops withdrawn prematurely, as so many Democrats seem to have been demanding lately. Again, imagine that.