[Posted by Karl]
RCP’s Sean Trende is not a big fan of forecasting, but today he revisits the model I mentioned in wondering whether there might be a “Romney effect” should Mitt Romney be the GOP nominee (the most likely outcome). Since he originally discussed this model, the campaign (more so the primaries than caucuses) have tended to show a geographic stability. Mitt has generally won in the Northeast and Mountain West, Santorum in the Great Plains, and Gingrich in the Deep South, with Romney and Santorum splitting the Midwest based on urban vs rural factors. The results have been fairly predictable, based on county shares of evangelicals, African-Americans, Latinos, college-educated voters, and Mormons. (These results support general findings from Jay Cost as well.)
Most troubling for Romney is that the more recent results show little to no momentum for Romney. This may or may not matter in a general election, but usually momentum does play a role in primaries. Trende is careful to note such momentum could develop. The most recent Rasmussen poll suggests Romney’s support has increased among Republicans, but that is not necessarily predicitive of voting in future primaries. If Mitt fails to get momentum — and if his rivals remain in the race — he will have a marginally worse slog than I anticipated in the upcoming Southern contests. Simply eyeballing the balance of evangelicals and Catholics, I thought Romney could have an opportunity in Louisiana, but the results of Trende’s model may suggest otherwise. Even so, I think the close states will remain close and that Indiana and Pennsylvania may not fall quite the way the model would suggest.