Presidential Polls: Closer Than You Might Think

Everywhere you turn, media organizations, political analysts, and casual observers in the blogosphere are citing polls and trying to predict who's winning in the race for the presidency. But for those who are anxious to believe what they are seeing in the poll results, and for those who deny what the poll results are telling them, here are a few words of caution.

Virtually every poll available right now has Democratic presidential nominee Barack Obama leading Republican nominee John McCain by a comfortable margin. The latest Real Clear Politics average, which looks at multiple polls instead of just one or two, has Obama leading McCain by nearly 8 points.

But it is not wise to look at results such as the RCP average and try to discern what will happen on November 4. Here's why.

First, all of the polls reported in the mainstream media are based on a two-person race. Respondents are asked to choose between Barack Obama and John McCain. But on election day, there will be more than two candidates on the ballot. And while the number of votes that will be cast for Libertarian Bob Barr and Independent Ralph Nader is negligible, those votes will still have an impact on the race. According to the Real Clear Politics site, when Barr and Nader are factored in, Obama's lead shrinks to just over 4 points. That's because Nader siphons off more votes from Obama than Barr does from McCain.

Second, most poll reporting is based strictly on national level responses. I personally have never put much stock in national levels polls during a presidential election because the race for the White House is ultimately a state-by-state contest. Simply put, no one wins the presidency just because they get the most total votes on election day. What matters most are the individual states.

Some states are solidly behind Obama, and some states are solidly behind McCain. That means that a handful of "battleground" states will end up determining the next president of the United States. Those "battleground" states are Ohio, Florida, Nevada, North Carolina, Missouri, Indiana, Colorado, West Virginia, and Virginia. A look at the Real Clear Politics averages for those states show that the presidential contest is anybody's game.

In Ohio, RCP shows Senator Obama ahead by less than three points, within the margin of error and clearly a toss-up. In Florida, Obama leads by nearly 4 points, outside the margin of error, but dangerously close in a vitally important state. In Nevada, Obama leads by 3.2 percent, and in North Carolina Obama leads by only 1.2 percent. Senator McCain is holding a razor thin lead in Missouri, currently less than 1 percent, and a more healthy 3.8 percent lead in Indiana. Obama leads in Colorado by 4 percentage points, and McCain leads in West Virginia by 2.2 percent. In Virginia, Senator Obama has widened his lead to 6.3 percentage points. RCP now has Virginia as "leaning Obama" instead of toss-up, but I include it here for a simple reason: Virginia has not gone for a Democratic presidential candidate since Lyndon Johnson, and the GOP is making a strong push in the Commonwealth in the final weeks.

So, we have 9 "battleground" states in play. Given a polling margin of error of three percentage points, Ohio, North Carolina, Missouri, and West Virginia are completely up in the air. Florida is barely outside the margin of error, as is Nevada. Indiana is leaning to McCain, but is by no means decided, and the same can be said of Colorado for Obama. In Virginia, Obama has surged of late, but Virginia political history suggests that this state, too, will come down to the wire.

Now, there's no question all the trend lines are moving in Obama's direction. He clearly has the momentum, and this has always been his race to lose. Given an unpopular president, the Iraq war, the current economic crisis, Afghanistan deteriorating, and the illusory promise of "change," Obama should be running away with this thing. But he's not. Obama's inability to close the door on the McCain candidacy, state-by-state polling that is much tighter than national polling, and the Bradley effect, the phenomenon where white voters tell pollsters they'll vote for the African-American candidate then change their minds in the privacy of the voting booth, will likely make this election a lot closer than the national polls are predicting.