Upset Brewing in Tennessee?

A week ago, the only public polls of the Tennessee Republican primary on Tuesday indicated Rick Santorum was headed for a big win, perhaps by 20 points, over Mitt Romney and Newt Gingrich. Now two new polls suggest the race has tightened considerably, and Romney is on the verge of scoring an upset win if the trend from the last few days continues in the state.

Coupled with victories in Maine, Arizona, Michigan, and Washington state in the last week, and expected victories in Massachusetts, Idaho, Vermont, and Virginia on Super Tuesday, Romney victories in two closely contested primaries in  Ohio and Tennessee would clearly restore his firm leadership position in the race for the nomination. In Ohio, the two most recent surveys both give Santorum a lead of 2%, within the margin of error.

A new Rasmussen survey shows Santorum with a 4-point lead (34% to 30%) in Tennessee. Newt Gingrich, who hails from neighboring Georgia, where he is expected to cruise to victory in the primary in that state on Tuesday, might have been expected to do well in Tennessee, but he trails badly in the Rasmussen poll and other surveys of the state.

Last week, a poll was released with general election matchups in Tennessee that showed GOP candidates ahead of Barack Obama, but by less than the margin of victory of recent GOP presidential nominees in the state. The state shifted to the GOP in 2000, when home-state Democrat Al Gore failed to carry the state (he would have been elected president had he held the Volunteer State). In 2004 and 2008, the state voted more heavily Republican, both for President Bush and then John McCain.

In the only deep South state to have voted so far in 2012, Newt Gingrich decisively beat Romney in South Carolina. So a win for Romney in Tennessee would do a lot to project his ability to win even in states that are viewed as conservative strongholds in a region where he was expected to do poorly.

Nate Silver, the New York Times political statistician, creates projection models for states. In primaries, where poll results are generally less reliable than in general elections, especially in the period a few weeks before the primary election, he values much more the most recent poll numbers. On Saturday, his model indicated that Santorum was expected to win Tennessee by 7 points, and that result was based largely on a YouGov survey that gave Santorum that exact margin, 37% to 30%. Earlier surveys in the state had Santorum up by 18-20 points, but Silver largely discarded those from his forecast methodology. His model has not yet been updated for the Rasmussen survey in Tennessee, but it is likely that inclusion of this latest poll might lower the expected margin of victory for Santorum, and also decrease his chance of winning the state, currently pegged by Silver at 84%.