Fun With Numbers -- Polls v. Votes

So we've got this lone GOP gubernatorial primary poll showing Terry Branstad with a 46-31 lead over Bob Vander Plaats. Rod Roberts gets 13 percent.

Since it broke Tuesday, this May poll by Public Policy Polling has prompted much dancing.

A downplaying Branstad  and The Iowa Republican's Craig Robinson smell a Democratic plot:

Public Policy Polling is a Democratic Party polling firm. Its clients include four members of Congress, all of whom are Democrats. The group also does polling pro-choice groups like NARAL and Planned Parenthood, labor unions like AFL-CIO and SEIU, and it also does polling for the North Carolina Democratic Party.

The question of why a North Carolina polling firm would conduct a poll in Iowa is as interesting as who the group is. The most likely scenario for why the out-of-state firm would all of the sudden poll in Iowa is because it could be connected to Iowans for Responsible Government, a shady 527 group that is headed up by liberal Democrats. That group is trying to influence the Republican primary for governor.

It should come as no surprise that, after almost a month of negative attacks on Terry Branstad, a poll was conducted to see how much damage was inflicted on his election efforts. There is no doubt that Iowans for Responsible Government's attacks have affected Branstad's polling. The relentless TV ads and direct mail campaign were bound to take a toll on Branstad's numbers.

Democrats have the same feeling about Rasmussen, btw.

Bleeding Heartland quotes the PPP pollster as saying plot schmot.

Still, plot or not, I don't think 46 percent in a hotly contested three-way primary is all that shabby.

Still, I understand the TB 5.0 discomfort. Races like this have a way of making pollsters look sort of foolish.

Consider this graph from a May 31, 2002, Gazette story four days before a three-way GOP gubernatorial primary:

A poll for KCCI-TV in Des Moines... this week of 601 likely voters shows (Doug) Gross with a 10-point lead, with 41 percent. Rep. Steve Sukup of Dougherty had 31 percent; Bob Vander Plaats of Sioux City, 17 percent; and 11 percent were undecided.

Here's what the actual primary results looked like:

Doug Gross, 71,478 (35.88 percent)

Steve Sukup 64,490 (32.37 percent)

Bob Vander Plaats 63,077 (31.66 percent)

Gross, despite his polling lead, barely avoided having the race thrown into a state convention. Sukup nearly won, and Vander Plaats did far better than polling suggested.

So uneasiness for Branstad is wise. But Gross says he doesn't think 2010 will be like 2002.

For one thing, back then, Gross said, Vander Plaats benefited from negative attacks firing back and forth between his camp and Sukup. BVP was the fresh, high-road candidate who collected late undecideds.

"I don't see the same kind of variability you saw in my race," Gross said. "And Branstad is a lot better candidate than I'd ever hope to be."

It's also interesting that the PPP has Vander Plaats at 31 percent, right where he was in 2002.

We'll know how accurate that poll is soon enough.