08/02/2008 05:12 am ET | Updated May 25, 2011

The Color Purple: An Examination of the 2008 Presidential Election in the Swing States

When it comes to presidential politics, in the words of the immortal Duke Ellington, "it don't mean a thing if it ain't got that swing." Because presidential elections are decided by the electoral vote and not the popular vote, and because every state except Nebraska and Maine awards all of its electoral votes to the candidate who wins a plurality of the popular vote, no matter how small the margin, presidential campaigns devote almost all of their resources on the swing states -- the minority of states where the race is close enough that both candidates believe they have a chance to win. In 2004, a dozen states fell into this category. Some were heavily populated states with large numbers of electoral votes such as Florida, Ohio, Michigan, and Pennsylvania. They received the lion's share of the campaigns' attention. But even less populous swing states with only a few electoral votes such as New Hampshire, Nevada, and New Mexico received visits from the candidates along with heavy doses of television advertising during the final weeks of the campaign.

Despite speculation by political commentators about the possibility of a dramatically different electoral map in 2008, it now appears that the outcome of this year's presidential race will come down to the same swing states as in 2004 along with a few additions. Once again Florida, Ohio, Michigan, and Pennsylvania are expected to receive a lot of attention from the presidential candidates. So are Iowa, Wisconsin, Minnesota, Oregon, Colorado, New Mexico, Nevada, and New Hampshire. This year, though, you can add Virginia, North Carolina, Missouri, Indiana, North Dakota, and Montana to the list of potential swing states based on the results of polls in the past few weeks.

If you live in one of the swing states, you can expect to be bombarded by positive and negative television ads. You may even get a chance to see John McCain or Barack Obama in person. But if you live in the rest of the country -- including populous states like New York, California, Texas, and Illinois with lots of electoral votes, you won't be seeing many ads and you probably won't get a chance to see either of the candidates unless you can afford to shell out several thousand dollars to attend a campaign fundraiser because that's the only reason they would be visiting your state.

So how are the candidates doing in the swing states? According to some recent polls and news stories, John McCain has been gaining ground on Barack Obama. A series of new Quinnipiac polls show McCain doing better than a few weeks ago in such key states as Michigan, Minnesota, Wisconsin, and Colorado. And a new Rasmussen poll shows McCain leading Obama by 10 points in Ohio after trailing him in most other recent polls.

These results, along with some national polls showing Obama with only a narrow lead, have led some political observers to speculate that despite extensive media coverage of Obama's trip to Europe and the Middle East and continuing dire economic news, the presidential race may be tightening, especially in the crucial swing states. But does the evidence actually support this conclusion?

The best place to look for evidence of any tightening of the presidential race in the swing states is the Gallup tracking poll. Gallup has been interviewing approximately 800 registered voters per night for several months and pooling these interviews into weekly samples to analyze trends in candidate preference in certain subgroups within the electorate. One of the trends Gallup has been tracking is support for John McCain and Barack Obama in the red, blue, and purple states.

Because the Gallup weekly samples involve more than 5000 interviews, the results can be considered highly reliable. In contrast, some state polls are based on samples of only 500 or 600 voters. The relatively small number of respondents along with differences in the sampling, interviewing, and weighting procedures used by different polling organizations can produce dramatically different results between polls conducted at approximately the same time in a particular state. For example, the Quinnipiac poll, based on interviews between July 14 and 22, showed Barack Obama leading John McCain by only 2 points in Minnesota. But a Rasmussen poll based on interviews conducted on July 22 showed Obama leading McCain by 13 points in Minnesota. Since the Quinnipiac poll was based on a much larger sample of likely voters (1261) than the Rasmussen poll (600), one would normally have more confidence in the Quinnipiac results. On the other hand, the results of the Rasmussen poll were much closer to the results of other recent polls in the state which might lead one to have more confidence in its results.

The dueling poll results in Minnesota are by no means unique. Within the past month different polling organizations have conducted surveys with widely differing results in Ohio (an 8 point Obama lead in a Public Policy Polling poll vs. a 10 point McCain lead in a Rasmussen poll), Florida (an 8 point McCain lead in a Strategic Vision poll vs. a 2 point Obama lead in a Rasmussen poll), and Missouri (a 5 point McCain lead in a Rasmussen poll vs. a 5 point Obama lead in a Research 2000 poll). These results suggest that it is hazardous to place too much weight on the results of individual state polls. Averaging the results of several polls in a state might be a reasonable approach, but there are rarely more than three or four polls conducted in any state in a single month.

In contrast to individual state polls, because of the size of the sample the weekly Gallup results should provide a reliable estimate of trends in candidate support in the purple states as well as the red and blue states. The following figure shows the trends over the past six weeks in Barack Obama's lead or deficit vs. John McCain in each set of states.


Source: Gallup Poll

The clear impression that one gets from the weekly Gallup tracking poll data is of a highly stable presidential race. Over six weeks of polling, Obama's lead over McCain in the blue states has ranged between 15 and 20 points while McCain's lead over Obama in the red states has ranged between 7 and 11 points. In the crucial purple states, Obama's lead over McCain has ranged from a low of 5 points to a high of 9 points. During the most recent week of polling Obama held a lead of 6 points.

There is no evidence in the weekly Gallup tracking poll data of any narrowing of the presidential race in the swing states. If we compare the results of the first three weeks of polling, from June 9 through June 29, with the results of the last three weeks of polling, from June 30 through July 20, we find that Obama's lead in the purple states averaged 7 points during the first three weeks and 8 points during the last three weeks. The results are similar in the blue states and red states. In the blue states, Obama's lead averaged 17 points during the first three weeks and 16 points during the last three weeks; in the red states, McCain's lead averaged 11 points during the first three weeks and 9 points during the last three weeks.

It is possible, of course, that there have been large but offsetting movements in different swing states in recent weeks. But this seems highly unlikely. If one of the candidates was making significant gains, we would expect to see those gains show up across the board. In fact, this is what happened between the first three months of the Gallup tracking poll, March through May, and the last two months, June and July. Between the first period and the second period, Obama increased his lead over McCain in the purple states from 2 points to 8 points. He also increased his lead in the blue states from 13 points to 16 points and reduced his deficit in the red states from 13 points to 10 points.

Since Barack Obama clinched the Democratic presidential nomination in early June the presidential race appears to have stabilized with Obama holding a modest but significant lead over John McCain in the swing states and in the nation as a whole. Nothing in the Gallup tracking poll data suggests that anything has changed recently.