Polling scribe Nate Silver dissects the recent results from the GOP's favorite pollster, Scott Rasmussen, and raises interesting questions about his methodology.

In this case, Rasmussen's latest shows that in the New York governor's race, Democrat Mario Cuomo enjoys a 16-point lead over Tea Party-backed candidate Carl Paladino. It's a big lead, although most political observers would probably put the race closer to the blowout range than does the Rasmussen poll.

But here's what's so interesting about the poll. When robo-calling respondents, Rasmussen did not include well-known Rick Lazio as the third candidate in the race even though Lazio has made it clear he plans to run all the way until November as the Conservative Party pick. (Paladino defeated Lazio in the GOP primary last week.) Silver argues that the outcome of the poll likely would have been different if Rasmussen including Lazio, who is expected to siphon off Paladino votes on Election Day.

And here's where it gets really interesting. Rasmussen on Sunday penned a pro-Tea Party opinion column in the New York Daily News scolding “pundits, political experts and commentators” for writing off Palandio's chances against Cuomo, insisting the race would soon tighten. And then what do you know, the next day Rasmussen releases polling results that show the race doing exactly that.

Amazing, right? Well, it is only if you forget to include Lazio in the poll.

Writes Silver:

Mr. Rasmussen is sometimes accused of wanting to “push a narrative” at the expense of what should be a pollster's goal, which is to reflect public opinion in the electorate he is testing as fairly and accurately as possible. Conducting this poll without including Mr. Lazio might not quiet his critics.

Another example of why it's not a good idea for a pollster to try to manufacture headlines. Yet Rasmussen seems to try to do that constantly.