A push poll is essentially a survey conducted by a pollster trying to get a fore-ordained conclusion. You do it by including disparaging information about one candidate in the questions. The process works for you in two ways: You can good poll numbers to crow about, and you’ve spread negative information about your opponent.

Last month, you’ll recall, a poll from Rasmussen, an established pollster, had John McCain and J.D. Hayworth running within a few points of each other in a potential matchup for the Republican senatorial primary next year.

The McCain organization, horrified at those numbers hanging over him, no doubt commissioned a more friendly poll, from a right-wing group called Tarrance, to combat it.

In the Republic the other day, Dan Nowicki did a story about it. Unsurprisingly, it showed McCain with a 20-point lede over a potential Hayworth challenge.

While Nowicki made it clear at the top it was a Republican poll, the hedline didn’t (in the print edition), and you had to read to the end of the article before you found out that the poll did smear Heyworth in the questions.

The pollster’s line is that McCain was winning before the questioners started disparaging Hayworth.

Still, I think the story should have led with the fact that it was a push poll. And Nowicki should have asked the pollster who specifically paid for it.

That said, he also didn’t report what to me was a big difference between the two polls. The Rasmussen one was a “robo poll” that questioned about 600 Arizonans via automated calling. The Tarrance one asked the same number by a live person over the phone. The latter is the superior process.