Naturally, a lot can be happen between now and November when we vote for President. Making early predictions are ripe with pitfalls where an incumbent President holds advantages that a challenger does not possess. There is always the possibility of an October surprise of some kind, we haven't even had the conventions yet and the inevitable bump in the polls the nominees receive, we have a presumptive nominee without a running mate, etc. In short, there is still a long way to go. With that in mind, ignoring the electoral college maps which anyone can play around with to create any scenario they want on websites like ElectionProjection, 270towin, or Realclearpolitics (to name three), I use a slightly different method.
Besides looking at the overall poll results, equally important are trends. These trends are factored in not only with respect to 2012 polls in each state where they are available, but also electoral trends over 11 Presidential election cycles and the last five Presidential election cycles. A score is derived for each state based on the deviation from the national win margin, much like a Cook PVI is determined. Unlike the Cook figures, these deviations over the 11 cycles indicate stability in voting patterns and help to discern recent trends which is why the the last 5 cycles are also examined to determine the difference between the 11 year average and the 5 year average. This lends itself to determining a trend score which serves more as a constant in the final equation. The 2012 poll trend analysis creates another "constant" that fluctuates from month to month, but is an indicator of trends specific to each state and, more broadly, momentum and/or the effect of the campaign message from either side. Using this analysis, for example, we can see that Oregon, although moving left, is not moving as dramatically left as a California. Conversely, it also indicates a drift rightward in Minnesota although the state is still too comfortably to the left to put in Romney's column. However, the current polling data and trend data can overcome that gap. And equally important, certain polls, based on past performance, are weighted more highly than others. SurveyUSA holds greater weight than a poll conducted by PPP or Rasmussen. I have also found that local polls- those conducted by entities within the state surveyed- are also slightly more accurate than national polling data. In this case, a poll in Virginia by a newspaper in Norfolk would hold equal or greater weight than even SurveyUSA or Mason-Dixon.
Obviously, Romney needs to win several states that Obama carried in 2008 in order to prevail. First, lets just forget Indiana which will be in the Romney column. This system also gives Romney razor thin victories in Virginia and Florida and a healthy victory in North Carolina. In fact, Romney should sweep the south.
Yet, that is still not enough to give Romney the Presidency and he would lose 290-248 in electoral votes. Romney needs to steal 22 electoral votes somewhere. The only single states that meet that criteria are New York (an unrealistic target) and Florida, who I already gave to Romney. That then means he really has to win two more states that Obama took in 2008. They would be: New Hampshire(4), Ohio (18), Iowa (6), Pennsylvania (20), Colorado (9) and Nevada (6). The only scenario where Romney can prevail is winning either Pennsylvania or Ohio, plus any of the other states listed.
Before going on, many believe that Michigan, Wisconsin and even Minnesota are in play and should be added to this list. However, the trend line in Presidential elections in Michigan has been towards the Democratic Party. In Minnesota, they have had the largest trend towards Republicans over the past five cycles, yet the gap is still so large that a Romney victory here is a long shot. And in Wisconsin, the trend line has been less than that in Minnesota yet still safely in the Democratic column. If Romney were to target any one of these three states, his best chance would be in Wisconsin.
So leaving aside the pipe dream of winning a state in the upper midwest, the attention has to turn to either Ohio or Pennsylvania. This analysis shows Ohio as being the more likely target. The trends in Pennsylvania essentially show stagnation with Obama prevailing there. However, in Ohio Romney trailed by an average of 8 points at the end of April to trailing by an average of 4 points at this point in May. If this trend holds in Ohio, then come Election Day, Romney has a real chance of gaining their 18 electoral votes. The best allocation of Romney resources would be in Ohio rather than Pennsylvania.
Yet, he still needs another 4 electoral votes to put him over the top. New Hampshire would do exactly that, but there are two other even more attainable selections. Some have mentioned Iowa, but Romney's chances there are about the same as they are in New Hampshire: close but no cigar. At the end of the day, Obama should squeak out a 3 point margin in New Hampshire, just as he did in 2008 while in Iowa the margin will be about 4% instead of the 5% he enjoyed in 2008.
The only two realistic states for Romney are either Nevada or Colorado. Of the two, Obama is most vulnerable in Nevada. In 2008, he managed a 3-point victory there compared to a 6-point victory in Colorado. My projections currently put Obama at a 1.08 point margin of victory in Nevada and 1.80 margin in Colorado. Both projected margins are close enough that Romney can exploit his way to victory. However, one has to look where those votes are going to come from in these states. That is, Romney needs to perform better than John McCain in certain categories in these states. With the traditional Republican demographic groups in these states, McCain performed about as well as Bush did with them in 2004. Some have pointed to the growing Hispanic populations in these states as a target demographic for Romney.
Focusing on the Hispanic vote exclusively and ignoring other demographic groups in these states would be a waste of time and money. If we were to just rely on Hispanic voters to tip the scales in these states, Romney would need 73% of the Hispanic vote in Colorado and 81% of the Hispanic vote in Nevada. Any Republican would be hard-pressed to split the Hispanic vote in either state. Because of the smaller population of Nevada, targeting groups to improve upon McCain 2008 performance incrementally would make better sense. If Romney can simply split the white vote, increase the Hispanic vote to Colorado levels in 2008 for the GOP and take in slightly more voters (about 15,000) nearing retirement, it would give him an 8,000 vote cushion and victory. A possible campaign that targets those nearing retirement (age 50-64) in Nevada by stressing the insolvency of Medicare and Social Security with another 4 years of Obama could be a winning strategy. All Romney needs in Nevada is another 15,000 votes from this demographic and an extra 45,000 Hispanics to go along with him and Nevada's 6 electoral votes would go to Romney giving him the Presidency.
In conclusion, it would appear at this point that Romney's path to victory includes (1) sweeping the south, (2) winning Ohio or Pennsylvania (most likely Ohio), and (3) winning Nevada. Assuming he achieves (1) and (2) then takes New Hampshire, then (3) is moot and we would know early on that Romney had won.