One would think that Gallup, Pew, Rasmussen, every sufficiently wealthy news organization and anyone else interested in conducting a poll would be familiar with the basics of the American electoral system. Why they all insist on continuing to waste precious ink on national polls, then, is completely mystifying.
Gallup's latest poll of registered voters reports that former Gov. Mitt Romney and President Barack Obama are tied nationally, 48 to 48 percent. Gallup's latest poll of likely voters, based on a complex set of assumptions about voter turnout, has Romney leading Obama by 5 percentage points, 51 to 46.
These figures are based on a national sample, so they theoretically include voters from Ohio, Florida and Virginia. They also include voters from Wyoming, California, Alabama, Delaware and about 40 other states whose voters could not possibly be any less relevant to the outcome on Nov. 6.
At this stage in the election, like any sufficiently close election, the fate of the candidates rests with fewer than a half-dozen states. The continuing snapshots of national polls are useful for pollsters and academics, who are interested in things like expected vote share or the probability of victory in the national popular vote. Most stakeholders care only about the likelihood of victory in the Electoral College, and a national poll is not very useful at this point.
This is why most prognosticators consider Obama to have a far higher chance of victory than the national polls would suggest. The Signal has Obama at a 65 percent chance of victory, while Nate Silver gives him a 75 percent chance against Romney. A small, demented chorus of observers has recently dinged Silver for this conclusion, citing various gut feelings to the contrary.
Odds of Obama victory vs. polls
Sources: Betfair, Intrade, IEM, HuffPost's Pollster and RealClearPolitics
Read More »from National polls are meaningless at this stage in the election