By Mark Mellman - 07/15/08 06:03 PM EDT
How big is big? The question arises in numerous contexts — here, in exploring the nature and significance of Barack Obama’s lead over John McCain.
As I have argued repeatedly, the fundamentals determine the outcome of presidential elections.
Never before in modern times have those fundamentals more clearly favored one party as they favor Democrats today. Today’s polls tell us much less about the ultimate outcome than do the fundamentals.
Nevertheless, I am fascinated by the continuing effort to “minimize” Obama’s lead in the polls. News organizations ought to be cautious in interpreting poll results and their bearing on elections, but many are doing statistical back flips to declare this race close. Witness a
recent CNN headline reporting on a 50 percent-45 percent Obama lead in their poll — “New CNN Poll: Obama, McCain in a statistical dead heat.”
Presumably, the “statistical dead heat” is a function of the poll’s 3.5 percentage-point margin of error, which makes it possible that McCain’s support is a few points higher and Obama’s a few points lower.
Possible, but not likely. Margin of error is at the same time both vastly underused in interpreting polls and vastly overused. Underused in that not every two-point shift in a tracking poll is meaningful; overused in that not every possibility that fits within the “margin of error” is equally likely. A few quick calculations reveal that in this CNN survey there is a 95 percent probability that Obama leads McCain. Perhaps not by 5 points, maybe by 2 or 3, but the odds are overwhelming that this poll is not a dead heat.
Add to that the fact that five other publicly released, conventional telephone polls were done during the same period and every single one of them showed Obama ahead by no fewer than four points. Together these polls represent a sample of nearly 9,000 interviews, which would have a margin of error of just 1 point.
Of course, statistical significance is only one measure of size. Context provides another. In the scheme of presidential elections, just how big would a four-point win be? Five of the last 12 presidential elections were decided by lesser margins; only one candidate in a contest that did not feature an incumbent won by more than four points. Thus, by any reasonable historical standard, should Obama sustain his four-point lead through November, it would be a substantial victory.
The Electoral College offers yet another metric for measuring the magnitude of a four-point margin. A popular-vote victory of that size implies winning about 310-325 electoral votes. Think about that for a moment. Winning the Kerry states plus Ohio would bring Obama to 272. Add Florida and he’s got 299. Getting to 325 would require adding both Ohio and Florida to the Kerry states, as well as Iowa, New Mexico, Nevada and Colorado. Of course, there is no way to know which particular states will fall. Perhaps Florida remains in the GOP column. Then getting to 325 electoral votes would mean adding, say, both Virginia and Georgia instead. By any construction, though, a four-point popular vote win means a sweeping victory in the Electoral College.
In the end, these polls are not so much close as malleable — and there is a difference. A four-point victory would not be close at all. However, there is no certainty that a four-point lead today will translate into a four-point lead in November. However, as Mike Dukakis discovered, that would also be true if Obama led by 17 points. While a close race may be a more interesting story, the simple truth is that, for now, this race isn’t that close.
Mellman is president of The Mellman Group and has worked for Democratic candidates and causes since 1982, including Sen. John Kerry (D-Mass.) in 2004.