Why Obama Could Be Winning by a Lot More Than You Think

Not everyone will agree, but for a number of reasons, this election seems like its Barack Obama’s to lose. He’s been routinely giving speeches to enthusiastic crowds of 50,000 or more, he’s been raising money faster than any other candidate in American history, and perhaps most importantly, he is a youthful, exciting, energetic candidate that represents a change in a change election.

John McCain on the other hand has struggled not only to keep pace financially, but also to inspire the same kind of enthusiasm in the electorate that Obama has demonstrated. With the faltering economy, an increasingly unpopular war, and close ties to an incumbent president with some of the lowest approval ratings in recorded history, it certainly seems like McCain would have his work cut out for him. In fact, all conventional wisdom says this should be a slam dunk win for Obama…

However, today’s Gallup tracking poll has Barack Obama listed with a mere three-point advantage over John McCain, 43% – 46%. Three points – that’s all? Can that be right? With everything that’s going on, how can it be that Obama’s only up 3 points in the polls?

In fact, most of the major tracking polls – Rasmussen, Newsweek, CNN, McLaughlin, – have the race at only a 1% – 5% point lead for Barack Obama; most are basically within the margin of error. How can it be that this seemingly lopsided race is so close? One shudders to think – but is it possible that all these polls are just plain wrong?

Maybe.

In 1948, President Harry Truman defeated Republican challenger Thomas A. Dewey by, what was at that time, the slimmest margin in the history of American politics. This narrow and frankly unexpected victory by Truman was forever immortalized by the photograph of Truman holding up the early edition of the November 3rd Chicago Tribune, bearing now-famous headline – “Dewey Defeats Truman”

How did the Chicago Tribune get it so wrong?

In the weeks before the election, the widely-read news magazine Literary Digest had conducted a national telephone poll that had shown Republican candidate Thomas A. Dewey with a substantial lead over the incumbent, Democrat Harry Truman. It was largely based on the results of this poll that the Chicago Tribune, (as well as several other reputable papers) decided to declare Dewey the early winner on the night of November 2nd.

What was not obvious at the time, however, was that the Literary Digest poll had failed to accurately capture a representative sample of the voting electorate. In 1948, telephones were relatively common in America, but there were still a small but significant percentage of voters without a telephone in the home. These Americans without telephones were typically much poorer than average and generally lived in the more rural areas of the country.

Though they typically vote Republican today, these poor, rural voters represented one of the most dependable voting blocks for the Democratic Party in the 1940’s. The most obvious effect of all this was that a very large percentage of people who intended to vote for President Truman all along were simply not being contacted by the pollsters. This is a phenomenon known as undersampling – basing a poll on a sample group that doesn’t accurately represent the entire population. When Election Day came, these poor, rural voters without telephones still went to the polls, and ultimately reelected President Truman.

“What does this have to do with the 2008 presidential election?” you might ask. The fact is, today’s political pollsters are struggling with a similar situation. Over the last 50 years or so, we have largely taken for granted the fact that nearly every household has at least one phone line associated with it. However, as the popularity and affordability of wireless phones has increased, this has become less and less the case.

According to a 2007 survey from the Center for Disease Control, as many 15.8% of American households are now considered “wireless only” – meaning that while they have cell phones, they have no longer have a traditional home phone, or “land-line.” Considering how pervasive wireless phones have become in our society, it makes a lot of sense; why would you pay extra for a home phone these days when almost everyone has a cell phone that goes everywhere with them and can be dependably contacted at almost any time? Slowly but surely, the home phone has been dying off.

This wouldn’t necessarily be a problem, except for the fact that federal law currently prevents pollsters from contacting individuals on their wireless phones. That means that as many as 15-20% of American voters are simply no longer being counted by today’s telephone polls.

That’s not all. This wireless sampling error tends to affect certain demographic groups much more heavily than others. Younger voters (18-35), college students, apartment renters, African-Americans, and Hispanics are all typically more likely to favor cell phones over home phones, and thus, less likely to be counted by any telephone poll. This is particularly important since most of these groups represent the key voting blocks of the Democratic Party, and in all likelihood, plan on voting for Barack Obama in the November election.

By contrast, those that are being consistently polled are typically older, more predominantly white, and far more likely to be homeowners and to have children. This second set of demographic groups tend to vote much more conservatively than the first – groups that favor the Republican Party, and in all likelihood plan on voting for John McCain.

So what happens when you consistently poll more heavily from the second group than you do from the first? The answer is that your polling results have the potential to become greatly skewed from reality – and in this case, skewed in favor of the Republican candidate. And while it’s certainly possible that 43% of those surveyed by Gallup favor John McCain, I’d bet a steak dinner that no where near that percentage of “wireless only” voters do.

We may be in for a big surprise this November 3, and if theory holds, the Democratic Party and Barack Obama may be the most direct benefactors. Due to the sampling errors related to wireless phone use, the 2008 election has the potential to become the largest polling embarrassment since “Dewey Defeats Truman” in 1948. It would certainly represent the ultimate irony if, just like Literary Digest did 60 years ago, the current polling results are dramatically upset because they neglected to take into account the number of Democrats in this country without home telephones.