Polling on Convention Eve

It’s been a rather eventful week in the polling arena, with various outfits (well-known and otherwise) conducting surveys and trying to get a sense of where the electorate stands on the eve of the Republican convention. Mitt Romney’s selection of Paul Ryan as his running mate resulted in anywhere from a 3 to 5% bump in the national polls for the GOP, but much of that has dissipated over the course of the week. Romney is running better in many state polls, and as pollsters move to a likely voter model, the conventional wisdom says that Romney’s numbers will eventually break through the 46-47% barrier he’s currently behind.

The week began with polls that seemed to defy gravity. An FMWB Florida poll showed Romney with a 14 point lead, mostly due to the polling firm’s making some, um, interesting assumptions about who will show up to vote in Florida on election day, for example, saying that only 2% of the electorate will be under 30. FMWB then came out with a Michigan poll showing Romney up 4, which sounds responsible, but the Florida poll made many people suspicious of any result from this pollster. Two days later, another Michigan poll had the president ahead. Strange doings.

Quinnipiac/CBS/NYT then entered the fray with swing state polls showing Obama ahead in Florida, Ohio and Wisconsin, but the polls seem to have overstated Democrats, which should give Romney hope. He’ll need at least two out of these three states if he hopes to win in November, though my sense is that if he wins in Wisconsin, then he’s going to win a lot more Midwestern states too. The University of Cincinnati supported the Ohio findings with their own poll, but they didn’t post party ID, so it’s difficult to see if their numbers were accurate.

Missouri also garnered attention this week, but for all the wrong reasons. Polling in the post-Akin comments era was promising at first, but two new polls, Rasmussen and Mason-Dixon show Claire McCaskill with a solid lead in the race. Curiously, Rasmussen released a poll showing Obama ahead by 1 in Missouri, but subsequent PPP and Mason-Dixon polls showed Romney with leads that I believe will stand through election day.

Another swing state, Colorado, was polled by Onsight Public Affairs and showed the president with a small lead, mostly due to strength among women and Latino voters. Romney will of course contest Colorado, but my sense is that this could be the first state he abandons if September polling shows that his money would be better spent in other markets, such as Pennsylvania, Wisconsin and Virginia.

By the end of the week, stasis had been restored. Obama claimed a smaller lead in the national polls, and was back where he started the week with the national trackers, Rasmussen +1 and in a tie according to Gallup. Still, it was a good week for Mitt Romney as his numbers slowly rose higher and his message began to gain traction. As he enters the GOP convention he can feel confident that he will gain a significant bump in the polls and emerge from Tampa with anywhere from a 5-7% lead in the polls, which would translate into a 7-10% gain. He’ll probably lose some of that lead after the Democratic convention, and we should see Romney with a small lead by the second week in September. Remember also that the latest jobs numbers will be released two days after the Democratic convention ends, and right now it doesn’t look as if those numbers will be promising for the president. This could lead to more of a bump for Romney.

The campaign is about to get underway in earnest after a summer of name-calling and side issues. Let the fun begin.