The big political news of the week thus far was a poll released yesterday by SurveyUSA that showed Ed Perlmutter leading Peggy Lamm by 20 points in their Democratic primary race in CD7 (Herb Rubenstein was a distant third in the poll). Whenever a poll is released during a campaign, the candidate on the low end of the stick always rushes to discredit it, and this case was no different. What was different was how completely against traditional wisdom this poll appeared to be — which is one reason I don’t believe it for a second.

First off, Perlmutter may very well be leading in this race. I don’t question that he could have the lead with one week to go, but there is no way that he is ahead by 20 points over Lamm. I just don’t see how that is plausible, and I would say the same thing if a poll showed Lamm up by 20 points. Both campaigns have spent a lot of time and money on this race, and for one candidate to be ahead by 20 points with one week to go just defies believability.

But if you’re going to believe me, then how did SurveyUSA come up with its numbers? How could they be so inaccurate?

The Lamm campaign will tell you that the methodology behind the poll was flawed, and to some degree they may be correct. But the bigger factor at play here is that the poll doesn’t seem to account for absentee ballot voters. By some accounts, as many as 12,000 people may have already voted in CD7 via absentee ballot, which could be more than half of the entire total come Aug. 8. About 20,000 people voted in the last Democratic primary in CD7, which was in 2002, and there won’t be more than 25,000 who vote this year. In short, by the time SurveyUSA conducted its poll, more than half of the people who were going to vote had probably already cast their ballot.

Now, it would be easy enough to poll absentee ballot voters — because the county provides records of who should have received their ballot in the mail — but most people are reluctant to tell someone on the telephone who they voted for. It’s one thing to tell a pollster who you think you might vote for, but it’s a lot harder to admit who you did vote for. Most people don’t even like to tell their friends who they voted for. It’s a private thing for many folks.

So here’s where the SurveyUSA poll is problematic: Does it show a lead for Perlmutter among all voters, or is it showing a big lead only among voters who haven’t already cast their ballot? Unless you know the answer to that question, the poll doesn’t mean much at all.