One meme getting some play this week is the idea that the military vote may be up for grabs. The best exposition of this I've seen is Benjamin Wallace-Wells's article in the Washington Monthly, Corps Voters, which is well worth a read.
Dramatic confirmation of this hypothesis appears to be found in this Army Times Presidential Poll. When I visited it, the numbers were:
Bush 31%
Clark 22%
Kucinich 16% [Shurley shome mishtake -ed.]
Dean 12%
Edwards 10%
Kerry 4%
Braun 2%
Sharpton 1%
Lieberman < 1%
Gephardt < 1%
If that were right it would mean that almost 70% would choose a Democrat over Bush. And that doesn't mean the conservative Lieberman, either.
But wait! This is not a scientific poll at all. By all appearances, it's just a tally of responses from one of those horrible unscientific 'polls' on the Army Times homepage. From that page there's no way to tell whether the 'poll' is limited to people from, say, .mil addresses, or whether we all get to vote. Nor is it clear how or whether the web site has anything in place to block repeat voting.
In other words, it's totally meaningless without some evidence that the voters are only servicepeople, and that they only get to vote once each. And even then, since the results are based on a self-selected sample, they would only be suggestive at best.
[I should maybe explain the “Shurley shome mishtake” comment is reference to a British joke of sorts, originating from the disreputable but funny Private Eye magazine.]
A trackback on the Dean blog suggests that the poll likely was freeped.
http://blogcgi.deanforamerica.com/cgi-bin/mt/mt-tb.cgi?__mode=view&entry_id=2037